diff --git a/docs/audioqc.ipynb b/docs/audioqc.ipynb new file mode 100644 index 00000000..f3b954da --- /dev/null +++ b/docs/audioqc.ipynb @@ -0,0 +1,823 @@ +{ + "cells": [ + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "# load derivatives\n", + "# randomly label\n", + "# feature selection\n", + "# train RFC\n", + "# train dual loop LOO\n" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Overall Training Process:\n", + "```\n", + "For each site (Outer Loop - Leave-One-Site-Out Cross-Validation):\n", + " Hold out one site as the test set\n", + " Train on the remaining N-1 sites\n", + "\n", + " For each hyperparameter configuration (Inner Loop - Grid/Random Search):\n", + " Select a combination of:\n", + " - Preprocessing method (normalization + feature selection)\n", + " - Model type (SVC-lin, SVC-rbf, RFC)\n", + " - Classifier hyperparameters (C, gamma, tree depth, etc.)\n", + "\n", + " For each fold in cross-validation (Inner Loop - Cross-Validation):\n", + " Split the training data into train/validation folds\n", + " Train the model on the training fold\n", + " Evaluate on the validation fold\n", + " \n", + " Select the best preprocessing/model configuration based on average validation performance\n", + "\n", + " Train the final model with the best hyperparameters & preprocessing on all N-1 training sites\n", + " Evaluate on the held-out site\n", + "\n", + "```" + ] + }, + { + "cell_type": "code", + "execution_count": 67, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "(12523, 133)\n", + "0 participant\n", + "1 task\n", + "2 F0semitoneFrom27.5Hz_sma3nz_amean\n", + "3 F0semitoneFrom27.5Hz_sma3nz_stddevNorm\n", + "4 F0semitoneFrom27.5Hz_sma3nz_percentile20.0\n", + "5 F0semitoneFrom27.5Hz_sma3nz_percentile50.0\n", + "6 F0semitoneFrom27.5Hz_sma3nz_percentile80.0\n", + "7 F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2\n", + "8 F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope\n", + "9 F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope\n", + "10 F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope\n", + "11 F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope\n", + "12 loudness_sma3_amean\n", + "13 loudness_sma3_stddevNorm\n", + "14 loudness_sma3_percentile20.0\n", + "15 loudness_sma3_percentile50.0\n", + "16 loudness_sma3_percentile80.0\n", + "17 loudness_sma3_pctlrange0-2\n", + "18 loudness_sma3_meanRisingSlope\n", + "19 loudness_sma3_stddevRisingSlope\n", + "20 loudness_sma3_meanFallingSlope\n", + "21 loudness_sma3_stddevFallingSlope\n", + "22 spectralFlux_sma3_amean\n", + "23 spectralFlux_sma3_stddevNorm\n", + "24 mfcc1_sma3_amean\n", + "25 mfcc1_sma3_stddevNorm\n", + "26 mfcc2_sma3_amean\n", + "27 mfcc2_sma3_stddevNorm\n", + "28 mfcc3_sma3_amean\n", + "29 mfcc3_sma3_stddevNorm\n", + "30 mfcc4_sma3_amean\n", + "31 mfcc4_sma3_stddevNorm\n", + "32 jitterLocal_sma3nz_amean\n", + "33 jitterLocal_sma3nz_stddevNorm\n", + "34 shimmerLocaldB_sma3nz_amean\n", + "35 shimmerLocaldB_sma3nz_stddevNorm\n", + "36 HNRdBACF_sma3nz_amean\n", + "37 HNRdBACF_sma3nz_stddevNorm\n", + "38 logRelF0-H1-H2_sma3nz_amean\n", + "39 logRelF0-H1-H2_sma3nz_stddevNorm\n", + "40 logRelF0-H1-A3_sma3nz_amean\n", + "41 logRelF0-H1-A3_sma3nz_stddevNorm\n", + "42 F1frequency_sma3nz_amean\n", + "43 F1frequency_sma3nz_stddevNorm\n", + "44 F1bandwidth_sma3nz_amean\n", + "45 F1bandwidth_sma3nz_stddevNorm\n", + "46 F1amplitudeLogRelF0_sma3nz_amean\n", + "47 F1amplitudeLogRelF0_sma3nz_stddevNorm\n", + "48 F2frequency_sma3nz_amean\n", + "49 F2frequency_sma3nz_stddevNorm\n", + "50 F2bandwidth_sma3nz_amean\n", + "51 F2bandwidth_sma3nz_stddevNorm\n", + "52 F2amplitudeLogRelF0_sma3nz_amean\n", + "53 F2amplitudeLogRelF0_sma3nz_stddevNorm\n", + "54 F3frequency_sma3nz_amean\n", + "55 F3frequency_sma3nz_stddevNorm\n", + "56 F3bandwidth_sma3nz_amean\n", + "57 F3bandwidth_sma3nz_stddevNorm\n", + "58 F3amplitudeLogRelF0_sma3nz_amean\n", + "59 F3amplitudeLogRelF0_sma3nz_stddevNorm\n", + "60 alphaRatioV_sma3nz_amean\n", + "61 alphaRatioV_sma3nz_stddevNorm\n", + "62 hammarbergIndexV_sma3nz_amean\n", + "63 hammarbergIndexV_sma3nz_stddevNorm\n", + "64 slopeV0-500_sma3nz_amean\n", + "65 slopeV0-500_sma3nz_stddevNorm\n", + "66 slopeV500-1500_sma3nz_amean\n", + "67 slopeV500-1500_sma3nz_stddevNorm\n", + "68 spectralFluxV_sma3nz_amean\n", + "69 spectralFluxV_sma3nz_stddevNorm\n", + "70 mfcc1V_sma3nz_amean\n", + "71 mfcc1V_sma3nz_stddevNorm\n", + "72 mfcc2V_sma3nz_amean\n", + "73 mfcc2V_sma3nz_stddevNorm\n", + "74 mfcc3V_sma3nz_amean\n", + "75 mfcc3V_sma3nz_stddevNorm\n", + "76 mfcc4V_sma3nz_amean\n", + "77 mfcc4V_sma3nz_stddevNorm\n", + "78 alphaRatioUV_sma3nz_amean\n", + "79 hammarbergIndexUV_sma3nz_amean\n", + "80 slopeUV0-500_sma3nz_amean\n", + "81 slopeUV500-1500_sma3nz_amean\n", + "82 spectralFluxUV_sma3nz_amean\n", + "83 loudnessPeaksPerSec\n", + "84 VoicedSegmentsPerSec\n", + "85 MeanVoicedSegmentLengthSec\n", + "86 StddevVoicedSegmentLengthSec\n", + "87 MeanUnvoicedSegmentLength\n", + "88 StddevUnvoicedSegmentLength\n", + "89 equivalentSoundLevel_dBp\n", + "90 duration\n", + "91 speaking_rate\n", + "92 articulation_rate\n", + "93 phonation_ratio\n", + "94 pause_rate\n", + "95 mean_pause_duration\n", + "96 mean_f0_hertz\n", + "97 std_f0_hertz\n", + "98 mean_intensity_db\n", + "99 std_intensity_db\n", + "100 range_ratio_intensity_db\n", + "101 mean_hnr_db\n", + "102 std_hnr_db\n", + "103 spectral_slope\n", + "104 spectral_tilt\n", + "105 cepstral_peak_prominence_mean\n", + "106 cepstral_peak_prominence_std\n", + "107 mean_f1_loc\n", + "108 std_f1_loc\n", + "109 mean_b1_loc\n", + "110 std_b1_loc\n", + "111 mean_f2_loc\n", + "112 std_f2_loc\n", + "113 mean_b2_loc\n", + "114 std_b2_loc\n", + "115 spectral_gravity\n", + "116 spectral_std_dev\n", + "117 spectral_skewness\n", + "118 spectral_kurtosis\n", + "119 local_jitter\n", + "120 localabsolute_jitter\n", + "121 rap_jitter\n", + "122 ppq5_jitter\n", + "123 ddp_jitter\n", + "124 local_shimmer\n", + "125 localDB_shimmer\n", + "126 apq3_shimmer\n", + "127 apq5_shimmer\n", + "128 apq11_shimmer\n", + "129 dda_shimmer\n", + "130 stoi\n", + "131 pesq\n", + "132 si_sdr\n" + ] + } + ], + "source": [ + "import pandas as pd\n", + "\n", + "# Load CSV file\n", + "features_df = pd.read_csv(\"/Users/isaacbevers/sensein/b2ai-wrapper/b2ai-data/bridge2ai-voice-corpus-3/derived/static_features.csv\") # Replace \"file.csv\" with your actual file path\n", + "\n", + "# Display the first few rows\n", + "print(features_df.shape)\n", + "\n", + "\n", + "for i, col in enumerate(features_features_df.columns):\n", + " print(i, col)" + ] + }, + { + "cell_type": "code", + "execution_count": 53, + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "(307, 1055)" + ] + }, + "execution_count": 53, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "phenotype_df = pd.read_csv(\"/Users/isaacbevers/sensein/b2ai-wrapper/b2ai-data/bridge2ai-voice-corpus-3/derived/phenotype.tsv\", sep='\\t')\n", + "phenotype_df.shape" + ] + }, + { + "cell_type": "code", + "execution_count": 59, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "(307, 1151)\n", + "True\n" + ] + }, + { + "data": { + "text/plain": [ + "Index(['record_id', 'redcap_repeat_instrument', 'redcap_repeat_instance',\n", + " 'selected_language', 'consent_status', 'is_feasibility_participant',\n", + " 'enrollment_institution', 'age', 'eligible_studies___1',\n", + " 'eligible_studies___2',\n", + " ...\n", + " 'vocabulary_item_word_4', 'vocabulary_item_difficulty_4',\n", + " 'vocabulary_item_word_5', 'vocabulary_item_difficulty_5',\n", + " 'vocabulary_item_word_6', 'vocabulary_item_difficulty_6',\n", + " 'random_session_id', 'random_recording_acoustic_task_id',\n", + " 'random_duration', 'random_item_generation_category'],\n", + " dtype='object', length=1151)" + ] + }, + "execution_count": 59, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "participants_df = pd.read_csv(\"/Users/isaacbevers/sensein/b2ai-wrapper/b2ai-data/bridge2ai-voice-corpus-3/bids/bids/participants.tsv\", sep=\"\\t\")\n", + "print(participants_df.shape)\n", + "print(\"session_site\" in participants_df.columns)\n", + "participants_df.columns" + ] + }, + { + "cell_type": "code", + "execution_count": 77, + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "(12523, 134)" + ] + }, + "execution_count": 77, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "participant_to_site = dict(zip(participants_df[\"record_id\"], participants_df[\"session_site\"]))\n", + "features_df[\"site\"] = features_df[\"participant\"].map(participant_to_site)\n", + "features_only_df = features_df.drop(columns=['site', 'participant', 'task']) # Exclude non-feature columns\n", + "features_df.shape" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### Preprocessing" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "> The first preprocessing step is a site-wise normalization of features. For robustness, this normalization calculates a center (as the median feature value) and a spread (as the interquartile range) per feature for demeaning and scaling data. This filter can center only, scale only or perform both centering and scaling." + ] + }, + { + "cell_type": "code", + "execution_count": 78, + "metadata": {}, + "outputs": [], + "source": [ + "# Site-wise normalization\n", + "import pandas as pd\n", + "import numpy as np\n", + "\n", + "def site_wise_normalization(features_df, features, site_column, mode=\"both\"):\n", + " \"\"\"\n", + " Perform site-wise normalization using median and interquartile range (IQR).\n", + " \n", + " Args:\n", + " features_df (pd.DataFrame): The dataset including site information.\n", + " features (pd.DataFrame): The feature columns to normalize.\n", + " site_column (str): The column representing site labels.\n", + " mode (str): 'center', 'scale', or 'both' (default).\n", + " \n", + " Returns:\n", + " pd.DataFrame: Normalized feature DataFrame.\n", + " \"\"\"\n", + " normalized_features = features.copy()\n", + "\n", + " for site in features_df[site_column].unique():\n", + " site_mask = features_df[site_column] == site\n", + " site_data = features[site_mask]\n", + "\n", + " median = site_data.median()\n", + " iqr = site_data.quantile(0.75) - site_data.quantile(0.25) # Interquartile Range (IQR)\n", + "\n", + " if mode == \"center\":\n", + " normalized_features.loc[site_mask] = site_data - median\n", + " elif mode == \"scale\":\n", + " normalized_features.loc[site_mask] = site_data / iqr\n", + " elif mode == \"both\":\n", + " normalized_features.loc[site_mask] = (site_data - median) / iqr\n", + "\n", + " return normalized_features\n", + "\n", + "# Define site column and feature columns\n", + "site_column = 'site'\n", + "features_normalized_center = site_wise_normalization(features_df, features_only_df, site_column='site', mode='center')\n", + "features_normalized_scale = site_wise_normalization(features_df, features_only_df, site_column='site', mode='scale')\n", + "features_normalized_both = site_wise_normalization(features_df, features_only_df, site_column='site', mode='both')\n" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + ">The second preprocessing step available is a dimensionality reduction filter excluding features highly predictive of the site of origin of data points. To do so, we fit a classifier based on extremely randomized trees [39], where the variables are the features and the responses are the sites of acquisition. We iteratively fit the classifier and remove the feature most predictive of the site at each step, until certain convergence criteria is met (either a maximum number of features to remove is reached or the performance of the classifier is very low and thus the remaining features do not predict the site at all)." + ] + }, + { + "cell_type": "code", + "execution_count": 74, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Initial site prediction accuracy: 0.79\n", + "Iteration 1: Removed 'mfcc2_sma3_amean', New Accuracy: 0.78\n", + "Iteration 2: Removed 'mfcc3_sma3_amean', New Accuracy: 0.78\n", + "Iteration 3: Removed 'mfcc2V_sma3nz_amean', New Accuracy: 0.77\n", + "Iteration 4: Removed 'slopeUV0-500_sma3nz_amean', New Accuracy: 0.76\n", + "Iteration 5: Removed 'mfcc1V_sma3nz_amean', New Accuracy: 0.75\n", + "Iteration 6: Removed 'slopeUV500-1500_sma3nz_amean', New Accuracy: 0.75\n", + "Iteration 7: Removed 'mfcc1_sma3_amean', New Accuracy: 0.74\n", + "Iteration 8: Removed 'mfcc3V_sma3nz_amean', New Accuracy: 0.73\n", + "Iteration 9: Removed 'loudness_sma3_percentile20.0', New Accuracy: 0.72\n", + "Iteration 10: Removed 'equivalentSoundLevel_dBp', New Accuracy: 0.72\n", + "Iteration 11: Removed 'alphaRatioUV_sma3nz_amean', New Accuracy: 0.71\n", + "Iteration 12: Removed 'hammarbergIndexUV_sma3nz_amean', New Accuracy: 0.71\n", + "Iteration 13: Removed 'spectralFlux_sma3_amean', New Accuracy: 0.71\n", + "Iteration 14: Removed 'spectralFluxUV_sma3nz_amean', New Accuracy: 0.71\n", + "Iteration 15: Removed 'mfcc4_sma3_amean', New Accuracy: 0.70\n", + "Iteration 16: Removed 'mfcc2_sma3_stddevNorm', New Accuracy: 0.69\n", + "Iteration 17: Removed 'loudness_sma3_amean', New Accuracy: 0.69\n", + "Iteration 18: Removed 'spectralFlux_sma3_stddevNorm', New Accuracy: 0.69\n", + "Iteration 19: Removed 'loudness_sma3_percentile80.0', New Accuracy: 0.69\n", + "Iteration 20: Removed 'loudness_sma3_pctlrange0-2', New Accuracy: 0.68\n", + "Iteration 21: Removed 'loudness_sma3_meanRisingSlope', New Accuracy: 0.69\n", + "Iteration 22: Removed 'loudness_sma3_stddevRisingSlope', New Accuracy: 0.68\n", + "Iteration 23: Removed 'loudness_sma3_percentile50.0', New Accuracy: 0.68\n", + "Iteration 24: Removed 'loudness_sma3_meanFallingSlope', New Accuracy: 0.67\n", + "Iteration 25: Removed 'loudness_sma3_stddevFallingSlope', New Accuracy: 0.67\n", + "Iteration 26: Removed 'mfcc3_sma3_stddevNorm', New Accuracy: 0.66\n", + "Iteration 27: Removed 'mfcc1_sma3_stddevNorm', New Accuracy: 0.66\n", + "Iteration 28: Removed 'loudness_sma3_stddevNorm', New Accuracy: 0.65\n", + "Iteration 29: Removed 'mfcc4_sma3_stddevNorm', New Accuracy: 0.65\n", + "Iteration 30: Removed 'loudnessPeaksPerSec', New Accuracy: 0.64\n", + "Iteration 31: Removed 'MeanUnvoicedSegmentLength', New Accuracy: 0.66\n", + "Iteration 32: Removed 'spectralFluxV_sma3nz_amean', New Accuracy: 0.65\n", + "Iteration 33: Removed 'slopeV0-500_sma3nz_amean', New Accuracy: 0.65\n", + "Iteration 34: Removed 'mfcc1V_sma3nz_stddevNorm', New Accuracy: 0.64\n", + "Iteration 35: Removed 'logRelF0-H1-A3_sma3nz_amean', New Accuracy: 0.64\n", + "Iteration 36: Removed 'hammarbergIndexV_sma3nz_amean', New Accuracy: 0.64\n", + "Iteration 37: Removed 'F0semitoneFrom27.5Hz_sma3nz_percentile20.0', New Accuracy: 0.63\n", + "Iteration 38: Removed 'F0semitoneFrom27.5Hz_sma3nz_percentile50.0', New Accuracy: 0.63\n", + "Iteration 39: Removed 'F0semitoneFrom27.5Hz_sma3nz_amean', New Accuracy: 0.62\n", + "Iteration 40: Removed 'F0semitoneFrom27.5Hz_sma3nz_percentile80.0', New Accuracy: 0.61\n", + "Iteration 41: Removed 'HNRdBACF_sma3nz_amean', New Accuracy: 0.61\n", + "Iteration 42: Removed 'logRelF0-H1-H2_sma3nz_amean', New Accuracy: 0.60\n", + "Iteration 43: Removed 'mfcc4V_sma3nz_amean', New Accuracy: 0.60\n", + "Iteration 44: Removed 'F1frequency_sma3nz_amean', New Accuracy: 0.60\n", + "Iteration 45: Removed 'F3frequency_sma3nz_amean', New Accuracy: 0.60\n", + "Iteration 46: Removed 'mfcc2V_sma3nz_stddevNorm', New Accuracy: 0.60\n", + "Iteration 47: Removed 'F2bandwidth_sma3nz_amean', New Accuracy: 0.60\n", + "Iteration 48: Removed 'slopeV500-1500_sma3nz_amean', New Accuracy: 0.59\n", + "Iteration 49: Removed 'F1frequency_sma3nz_stddevNorm', New Accuracy: 0.59\n", + "Iteration 50: Removed 'shimmerLocaldB_sma3nz_stddevNorm', New Accuracy: 0.58\n", + "Iteration 51: Removed 'F2frequency_sma3nz_amean', New Accuracy: 0.58\n", + "Iteration 52: Removed 'alphaRatioV_sma3nz_amean', New Accuracy: 0.58\n", + "Iteration 53: Removed 'F1bandwidth_sma3nz_amean', New Accuracy: 0.58\n", + "Iteration 54: Removed 'F3bandwidth_sma3nz_amean', New Accuracy: 0.58\n", + "Iteration 55: Removed 'shimmerLocaldB_sma3nz_amean', New Accuracy: 0.57\n", + "Iteration 56: Removed 'VoicedSegmentsPerSec', New Accuracy: 0.57\n", + "Iteration 57: Removed 'StddevUnvoicedSegmentLength', New Accuracy: 0.57\n", + "Iteration 58: Removed 'mfcc3V_sma3nz_stddevNorm', New Accuracy: 0.57\n", + "Iteration 59: Removed 'F2bandwidth_sma3nz_stddevNorm', New Accuracy: 0.57\n", + "Iteration 60: Removed 'F0semitoneFrom27.5Hz_sma3nz_stddevNorm', New Accuracy: 0.57\n", + "Iteration 61: Removed 'HNRdBACF_sma3nz_stddevNorm', New Accuracy: 0.56\n", + "Iteration 62: Removed 'F2amplitudeLogRelF0_sma3nz_amean', New Accuracy: 0.56\n", + "Iteration 63: Removed 'F3amplitudeLogRelF0_sma3nz_amean', New Accuracy: 0.57\n", + "Iteration 64: Removed 'F1amplitudeLogRelF0_sma3nz_amean', New Accuracy: 0.56\n", + "Iteration 65: Removed 'F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2', New Accuracy: 0.57\n", + "Iteration 66: Removed 'F2frequency_sma3nz_stddevNorm', New Accuracy: 0.56\n", + "Iteration 67: Removed 'F3amplitudeLogRelF0_sma3nz_stddevNorm', New Accuracy: 0.56\n", + "Final removed features: ['mfcc2_sma3_amean', 'mfcc3_sma3_amean', 'mfcc2V_sma3nz_amean', 'slopeUV0-500_sma3nz_amean', 'mfcc1V_sma3nz_amean', 'slopeUV500-1500_sma3nz_amean', 'mfcc1_sma3_amean', 'mfcc3V_sma3nz_amean', 'loudness_sma3_percentile20.0', 'equivalentSoundLevel_dBp', 'alphaRatioUV_sma3nz_amean', 'hammarbergIndexUV_sma3nz_amean', 'spectralFlux_sma3_amean', 'spectralFluxUV_sma3nz_amean', 'mfcc4_sma3_amean', 'mfcc2_sma3_stddevNorm', 'loudness_sma3_amean', 'spectralFlux_sma3_stddevNorm', 'loudness_sma3_percentile80.0', 'loudness_sma3_pctlrange0-2', 'loudness_sma3_meanRisingSlope', 'loudness_sma3_stddevRisingSlope', 'loudness_sma3_percentile50.0', 'loudness_sma3_meanFallingSlope', 'loudness_sma3_stddevFallingSlope', 'mfcc3_sma3_stddevNorm', 'mfcc1_sma3_stddevNorm', 'loudness_sma3_stddevNorm', 'mfcc4_sma3_stddevNorm', 'loudnessPeaksPerSec', 'MeanUnvoicedSegmentLength', 'spectralFluxV_sma3nz_amean', 'slopeV0-500_sma3nz_amean', 'mfcc1V_sma3nz_stddevNorm', 'logRelF0-H1-A3_sma3nz_amean', 'hammarbergIndexV_sma3nz_amean', 'F0semitoneFrom27.5Hz_sma3nz_percentile20.0', 'F0semitoneFrom27.5Hz_sma3nz_percentile50.0', 'F0semitoneFrom27.5Hz_sma3nz_amean', 'F0semitoneFrom27.5Hz_sma3nz_percentile80.0', 'HNRdBACF_sma3nz_amean', 'logRelF0-H1-H2_sma3nz_amean', 'mfcc4V_sma3nz_amean', 'F1frequency_sma3nz_amean', 'F3frequency_sma3nz_amean', 'mfcc2V_sma3nz_stddevNorm', 'F2bandwidth_sma3nz_amean', 'slopeV500-1500_sma3nz_amean', 'F1frequency_sma3nz_stddevNorm', 'shimmerLocaldB_sma3nz_stddevNorm', 'F2frequency_sma3nz_amean', 'alphaRatioV_sma3nz_amean', 'F1bandwidth_sma3nz_amean', 'F3bandwidth_sma3nz_amean', 'shimmerLocaldB_sma3nz_amean', 'VoicedSegmentsPerSec', 'StddevUnvoicedSegmentLength', 'mfcc3V_sma3nz_stddevNorm', 'F2bandwidth_sma3nz_stddevNorm', 'F0semitoneFrom27.5Hz_sma3nz_stddevNorm', 'HNRdBACF_sma3nz_stddevNorm', 'F2amplitudeLogRelF0_sma3nz_amean', 'F3amplitudeLogRelF0_sma3nz_amean', 'F1amplitudeLogRelF0_sma3nz_amean', 'F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2', 'F2frequency_sma3nz_stddevNorm', 'F3amplitudeLogRelF0_sma3nz_stddevNorm']\n", + "Final site prediction accuracy: 0.56\n" + ] + } + ], + "source": [ + "import pandas as pd\n", + "from sklearn.ensemble import ExtraTreesClassifier\n", + "from sklearn.model_selection import train_test_split\n", + "\n", + "X = features_df.drop(columns=['site', 'participant', 'task']) # Drop non-AQM columns\n", + "y = features_df['site']\n", + "\n", + "# Split data into train and test sets\n", + "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, stratify=y, random_state=42)\n", + "\n", + "# Train initial ExtraTreesClassifier to predict site labels\n", + "site_predictor = ExtraTreesClassifier(n_estimators=100, random_state=42)\n", + "site_predictor.fit(X_train, y_train)\n", + "\n", + "# Get initial accuracy (baseline)\n", + "initial_accuracy = site_predictor.score(X_test, y_test)\n", + "print(f\"Initial site prediction accuracy: {initial_accuracy:.2f}\")\n", + "\n", + "# Define stopping criteria\n", + "num_sites = len(np.unique(y)) # Number of unique sites\n", + "chance_level = 1 / num_sites # Random guessing accuracy\n", + "max_removals = 131 - 64 # Set a maximum number of features to remove (total features - MRIqc number)\n", + "\n", + "# Start feature elimination loop\n", + "features_to_remove = []\n", + "iteration = 0\n", + "\n", + "while iteration < max_removals:\n", + " # Get feature importances\n", + " feature_importances = site_predictor.feature_importances_\n", + " \n", + " # Identify the most predictive feature\n", + " most_predictive_feature = X_train.columns[np.argmax(feature_importances)]\n", + " features_to_remove.append(most_predictive_feature)\n", + "\n", + " # Remove the most predictive feature from the dataset\n", + " X_train = X_train.drop(columns=[most_predictive_feature])\n", + " X_test = X_test.drop(columns=[most_predictive_feature])\n", + "\n", + " # Retrain the classifier without the removed feature\n", + " site_predictor = ExtraTreesClassifier(n_estimators=100, random_state=42)\n", + " site_predictor.fit(X_train, y_train)\n", + "\n", + " # Get new accuracy\n", + " new_accuracy = site_predictor.score(X_test, y_test)\n", + " print(f\"Iteration {iteration + 1}: Removed '{most_predictive_feature}', New Accuracy: {new_accuracy:.2f}\")\n", + "\n", + " # Check stopping conditions\n", + " if new_accuracy <= chance_level:\n", + " print(\"Stopping: Site prediction accuracy is near chance level.\")\n", + " break\n", + "\n", + " iteration += 1\n", + "\n", + "print(f\"Final removed features: {features_to_remove}\")\n", + "print(f\"Final site prediction accuracy: {new_accuracy:.2f}\")\n" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "> Finally, a third preprocessing step implements the Winnow algorithm [40] using extremely randomized trees in a similar way to the previous filter, but comparing features to a synthetic, randomly-generated feature. This feature selection filter removes those IQMs below a certain SNR level." + ] + }, + { + "cell_type": "code", + "execution_count": 82, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Removed 35 low-SNR features.\n", + "Removed 32 low-SNR features.\n", + "Removed 38 low-SNR features.\n" + ] + } + ], + "source": [ + "import numpy as np\n", + "import pandas as pd\n", + "from sklearn.ensemble import ExtraTreesClassifier\n", + "\n", + "def winnow_feature_selection(X, y, snr_threshold=1.0):\n", + " \"\"\"\n", + " Perform Winnow-based feature selection using ExtraTreesClassifier.\n", + " \n", + " Args:\n", + " X (pd.DataFrame): Feature matrix.\n", + " y (pd.Series): Target labels (e.g., site).\n", + " snr_threshold (float): Minimum SNR threshold for feature retention.\n", + " \n", + " Returns:\n", + " pd.DataFrame: Reduced feature set with low SNR features removed.\n", + " \"\"\"\n", + " X = X.copy()\n", + " \n", + " # Generate a synthetic random feature (noise)\n", + " np.random.seed(42)\n", + " X[\"random_noise\"] = np.random.normal(0, 1, size=len(X))\n", + "\n", + " # Train ExtraTreesClassifier to measure feature importance\n", + " clf = ExtraTreesClassifier(n_estimators=100, random_state=42)\n", + " clf.fit(X, y)\n", + " \n", + " # Get feature importances\n", + " feature_importances = clf.feature_importances_\n", + " feature_names = X.columns\n", + " \n", + " # Identify the importance of the synthetic feature (random noise)\n", + " random_feature_importance = feature_importances[X.columns.get_loc(\"random_noise\")]\n", + "\n", + " # Compute SNR for each feature (ratio of feature importance to random noise importance)\n", + " snr = feature_importances / random_feature_importance\n", + "\n", + " # Select features where SNR exceeds the threshold\n", + " selected_features = feature_names[snr > snr_threshold].tolist()\n", + "\n", + " # Avoid ValueError: Only remove \"random_noise\" if it exists\n", + " if \"random_noise\" in selected_features:\n", + " selected_features.remove(\"random_noise\")\n", + "\n", + " print(f\"Removed {X.shape[1] - len(selected_features)} low-SNR features.\")\n", + " \n", + " return X[selected_features]\n", + "\n", + "\n", + "# Apply Winnow feature selection for each normalization method\n", + "features_selected_center = winnow_feature_selection(features_normalized_center, features_df[\"site\"])\n", + "features_selected_scale = winnow_feature_selection(features_normalized_scale, features_df[\"site\"])\n", + "features_selected_both = winnow_feature_selection(features_normalized_both, features_df[\"site\"])\n" + ] + }, + { + "cell_type": "code", + "execution_count": 84, + "metadata": {}, + "outputs": [], + "source": [ + "def add_labels(df, column_name=\"label\", labels=[\"accept\", \"exclude\", \"unsure\"], random_seed=None):\n", + " \"\"\"\n", + " Adds a column with randomly assigned labels to the DataFrame.\n", + "\n", + " Parameters:\n", + " - df (pd.DataFrame): The DataFrame to modify.\n", + " - column_name (str): Name of the new column (default: \"label\").\n", + " - labels (list): List of labels to sample from (default: [\"accept\", \"exclude\", \"unsure\"]).\n", + " - random_seed (int or None): Random seed for reproducibility (default: None).\n", + "\n", + " Returns:\n", + " - pd.DataFrame: Modified DataFrame with the new column.\n", + " \"\"\"\n", + " if random_seed is not None:\n", + " np.random.seed(random_seed) # Set random seed for reproducibility\n", + "\n", + " df[column_name] = np.random.choice(labels, size=len(df))\n", + " return df\n", + "\n", + "features_selected_center = add_labels(features_selected_center)\n", + "features_selected_scale = add_labels(features_selected_scale)\n", + "features_normalized_both = add_labels(features_normalized_both)\n", + "\n" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + ">A support-vector machine [37] finds a hyperplane in the high-dimensional space of the features that robustly separates the classes of interest. The SVC then uses the hyperplane to decide the class that is assigned to new samples in the space of features. Two hyper-parameters define the support-vector machine algorithm: a kernel function that defines the similarity between data points to ultimately compute a distance to the hyperplane, and a regularization weight C. In particular, we analyzed here the linear SVC implementation (as of now, “SVC-lin”) and the one based on radial basis functions (denoted by “SVC-rbf”). During model selection, we evaluated the regularization weight C of both SVC and the γ parameter (kernel width) particular to the SVC-rbf." + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Warning: NaN values detected in features. Imputing missing values...\n", + "Best Linear SVC (C=10), Test Accuracy: 0.3269\n", + "Best RBF SVC (C=0.1, gamma=1), Test Accuracy: 0.3397\n", + "Warning: NaN values detected in features. Imputing missing values...\n" + ] + } + ], + "source": [ + "import numpy as np\n", + "import pandas as pd\n", + "from sklearn.svm import SVC\n", + "from sklearn.model_selection import train_test_split, GridSearchCV\n", + "from sklearn.preprocessing import StandardScaler\n", + "from sklearn.metrics import accuracy_score\n", + "from sklearn.impute import SimpleImputer\n", + "\n", + "def train_svm_for_label_prediction(features_df, label_column=\"label\", test_size=0.2, random_state=42):\n", + " \"\"\"\n", + " Trains Support Vector Machines (SVM) with hyperparameter tuning to predict labels ('accept', 'exclude', 'unsure').\n", + "\n", + " Parameters:\n", + " - features_df (pd.DataFrame): DataFrame containing features and the label column.\n", + " - label_column (str): Name of the column containing classification labels (default: \"label\").\n", + " - test_size (float): Fraction of data to use for testing (default: 0.2).\n", + " - random_state (int): Random seed for reproducibility.\n", + "\n", + " Returns:\n", + " - dict: Contains best models and their test accuracies.\n", + " \"\"\"\n", + "\n", + " # Ensure label column exists\n", + " if label_column not in features_df.columns:\n", + " raise ValueError(f\"Label column '{label_column}' not found in DataFrame.\")\n", + "\n", + " # Separate features and labels\n", + " X = features_df.drop(columns=[label_column]) # Features\n", + " y = features_df[label_column] # Labels\n", + "\n", + " # Check for NaN values\n", + " if X.isna().sum().sum() > 0:\n", + " print(\"Warning: NaN values detected in features. Imputing missing values...\")\n", + " imputer = SimpleImputer(strategy=\"mean\") # Fill NaN with mean of the column\n", + " X = pd.DataFrame(imputer.fit_transform(X), columns=X.columns)\n", + "\n", + " # Split dataset into train and test sets\n", + " X_train, X_test, y_train, y_test = train_test_split(\n", + " X, y, test_size=test_size, stratify=y, random_state=random_state\n", + " )\n", + "\n", + " # Standardize features (SVM performs better with standardized inputs)\n", + " scaler = StandardScaler()\n", + " X_train_scaled = scaler.fit_transform(X_train)\n", + " X_test_scaled = scaler.transform(X_test)\n", + "\n", + " # Define hyperparameter grids for linear SVC and RBF SVC\n", + " param_grid_lin = {'C': [0.1, 1, 10, 100]}\n", + " param_grid_rbf = {'C': [0.1, 1, 10, 100], 'gamma': [0.001, 0.01, 0.1, 1]}\n", + "\n", + " # Perform Grid Search for Linear SVC\n", + " grid_search_lin = GridSearchCV(SVC(kernel=\"linear\"), param_grid_lin, cv=5, scoring=\"accuracy\")\n", + " grid_search_lin.fit(X_train_scaled, y_train)\n", + " best_svc_lin = grid_search_lin.best_estimator_\n", + "\n", + " # Perform Grid Search for RBF SVC\n", + " grid_search_rbf = GridSearchCV(SVC(kernel=\"rbf\"), param_grid_rbf, cv=5, scoring=\"accuracy\")\n", + " grid_search_rbf.fit(X_train_scaled, y_train)\n", + " best_svc_rbf = grid_search_rbf.best_estimator_\n", + "\n", + " # Evaluate the best models on test set\n", + " y_pred_lin = best_svc_lin.predict(X_test_scaled)\n", + " y_pred_rbf = best_svc_rbf.predict(X_test_scaled)\n", + "\n", + " acc_lin = accuracy_score(y_test, y_pred_lin)\n", + " acc_rbf = accuracy_score(y_test, y_pred_rbf)\n", + "\n", + " # Print results\n", + " print(f\"Best Linear SVC (C={grid_search_lin.best_params_['C']}), Test Accuracy: {acc_lin:.4f}\")\n", + " print(f\"Best RBF SVC (C={grid_search_rbf.best_params_['C']}, gamma={grid_search_rbf.best_params_['gamma']}), Test Accuracy: {acc_rbf:.4f}\")\n", + "\n", + " # Return best models and accuracies\n", + " return {\n", + " \"best_linear_svc\": best_svc_lin,\n", + " \"best_rbf_svc\": best_svc_rbf,\n", + " \"accuracy_linear\": acc_lin,\n", + " \"accuracy_rbf\": acc_rbf\n", + " }\n", + "\n", + "# Example Usage on Labeled Feature Sets\n", + "results_center = train_svm_for_label_prediction(features_selected_center)\n", + "results_scale = train_svm_for_label_prediction(features_selected_scale)\n", + "results_both = train_svm_for_label_prediction(features_normalized_both)\n" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "import numpy as np\n", + "import pandas as pd\n", + "from sklearn.ensemble import RandomForestClassifier\n", + "from sklearn.model_selection import train_test_split, GridSearchCV\n", + "from sklearn.preprocessing import StandardScaler\n", + "from sklearn.metrics import accuracy_score\n", + "from sklearn.impute import SimpleImputer\n", + "\n", + "def train_random_forest(features_df, label_column=\"label\", test_size=0.2, random_state=42):\n", + " \"\"\"\n", + " Trains a Random Forest Classifier (RFC) with hyperparameter tuning using GridSearchCV.\n", + "\n", + " Parameters:\n", + " - features_df (pd.DataFrame): DataFrame containing features and the label column.\n", + " - label_column (str): Name of the column containing classification labels (default: \"label\").\n", + " - test_size (float): Fraction of data to use for testing (default: 0.2).\n", + " - random_state (int): Random seed for reproducibility.\n", + "\n", + " Returns:\n", + " - dict: Contains best model and its test accuracy.\n", + " \"\"\"\n", + "\n", + " # Ensure label column exists\n", + " if label_column not in features_df.columns:\n", + " raise ValueError(f\"Label column '{label_column}' not found in DataFrame.\")\n", + "\n", + " # Separate features and labels\n", + " X = features_df.drop(columns=[label_column]) # Features\n", + " y = features_df[label_column] # Labels\n", + "\n", + " # Check for NaN values and impute missing data\n", + " if X.isna().sum().sum() > 0:\n", + " print(\"Warning: NaN values detected in features. Imputing missing values...\")\n", + " imputer = SimpleImputer(strategy=\"mean\") # Fill NaN with mean of the column\n", + " X = pd.DataFrame(imputer.fit_transform(X), columns=X.columns)\n", + "\n", + " # Split dataset into train and test sets\n", + " X_train, X_test, y_train, y_test = train_test_split(\n", + " X, y, test_size=test_size, stratify=y, random_state=random_state\n", + " )\n", + "\n", + " # Standardize features (optional for tree-based models, but helps when combined with other models)\n", + " scaler = StandardScaler()\n", + " X_train_scaled = scaler.fit_transform(X_train)\n", + " X_test_scaled = scaler.transform(X_test)\n", + "\n", + " # Define hyperparameter grid for Random Forest\n", + " param_grid = {\n", + " \"n_estimators\": [50, 100, 200], # Number of trees\n", + " \"max_depth\": [None, 10, 20], # Maximum depth of trees\n", + " \"min_samples_split\": [2, 5, 10], # Minimum samples required to split an internal node\n", + " \"min_samples_leaf\": [1, 2, 4], # Minimum number of samples required to be at a leaf node\n", + " }\n", + "\n", + " # Perform Grid Search for Random Forest\n", + " grid_search = GridSearchCV(RandomForestClassifier(random_state=random_state), param_grid, cv=5, scoring=\"accuracy\", n_jobs=-1)\n", + " grid_search.fit(X_train_scaled, y_train)\n", + " best_rfc = grid_search.best_estimator_\n", + "\n", + " # Evaluate the best model on test set\n", + " y_pred = best_rfc.predict(X_test_scaled)\n", + " acc = accuracy_score(y_test, y_pred)\n", + "\n", + " # Print results\n", + " print(f\"Best Random Forest Model: {grid_search.best_params_}, Test Accuracy: {acc:.4f}\")\n", + "\n", + " # Return best model and accuracy\n", + " return {\n", + " \"best_rfc\": best_rfc,\n", + " \"accuracy\": acc\n", + " }\n", + "\n", + "# Example Usage on Labeled Feature Sets\n", + "results_center = train_random_forest(features_selected_center)\n", + "results_scale = train_random_forest(features_selected_scale)\n", + "results_both = train_random_forest(features_normalized_both)\n" + ] + } + ], + "metadata": { + "kernelspec": { + "display_name": "b2ai", + "language": "python", + "name": "python3" + }, + "language_info": { + "codemirror_mode": { + "name": "ipython", + "version": 3 + }, + "file_extension": ".py", + "mimetype": "text/x-python", + "name": "python", + "nbconvert_exporter": "python", + "pygments_lexer": "ipython3", + "version": "3.11.11" + } + }, + "nbformat": 4, + "nbformat_minor": 2 +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/final_model/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/final_model/model.joblib new file mode 100644 index 00000000..2f3efea2 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/final_model/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/final_model/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/final_model/training_metadata.json new file mode 100644 index 00000000..05900fcc --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/final_model/training_metadata.json @@ -0,0 +1,94 @@ +{ + "best_preprocessing_steps": [ + [ + "eliminate", + "normalize" + ], + "scale" + ], + "best_model_score": 0.55, + "model_type": "SVC", + "hyperparameters": { + "C": 1, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + }, + "selected_features": [ + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_stddevNorm", + "mfcc1V_sma3nz_stddevNorm", + "MeanVoicedSegmentLengthSec", + "StddevUnvoicedSegmentLength", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" + ] +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate/RandomForest/model.joblib new file mode 100644 index 00000000..a1eeee59 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate/RandomForest/training_metadata.json new file mode 100644 index 00000000..e387c1f6 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate/RandomForest/training_metadata.json @@ -0,0 +1,29 @@ +{ + "preprocessing_steps": [ + "eliminate" + ], + "mode": null, + "cv_score": 0.3, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate/SVM/model.joblib new file mode 100644 index 00000000..239c9a18 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate/SVM/training_metadata.json new file mode 100644 index 00000000..ee5d60ca --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate/SVM/training_metadata.json @@ -0,0 +1,25 @@ +{ + "preprocessing_steps": [ + "eliminate" + ], + "mode": null, + "cv_score": 0.475, + "model_type": "SVC", + "hyperparameters": { + "C": 1, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate/selected_features.json new file mode 100644 index 00000000..7858a341 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate/selected_features.json @@ -0,0 +1,66 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "shimmerLocaldB_sma3nz_amean", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_stddevNorm", + "mfcc1V_sma3nz_stddevNorm", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "StddevUnvoicedSegmentLength", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_normalize/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_normalize/RandomForest/model.joblib new file mode 100644 index 00000000..114431a3 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_normalize/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_normalize/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_normalize/RandomForest/training_metadata.json new file mode 100644 index 00000000..d0145395 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_normalize/RandomForest/training_metadata.json @@ -0,0 +1,30 @@ +{ + "preprocessing_steps": [ + "eliminate", + "normalize" + ], + "mode": "both", + "cv_score": 0.35, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_normalize/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_normalize/SVM/model.joblib new file mode 100644 index 00000000..781699f0 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_normalize/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_normalize/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_normalize/SVM/training_metadata.json new file mode 100644 index 00000000..3285a8b5 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_normalize/SVM/training_metadata.json @@ -0,0 +1,26 @@ +{ + "preprocessing_steps": [ + "eliminate", + "normalize" + ], + "mode": "both", + "cv_score": 0.475, + "model_type": "SVC", + "hyperparameters": { + "C": 1, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_normalize/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_normalize/selected_features.json new file mode 100644 index 00000000..ae9e8e91 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_normalize/selected_features.json @@ -0,0 +1,66 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "logRelF0-H1-H2_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "mfcc1V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "VoicedSegmentsPerSec", + "StddevVoicedSegmentLengthSec", + "StddevUnvoicedSegmentLength", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_normalize_winnow/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_normalize_winnow/RandomForest/model.joblib new file mode 100644 index 00000000..ad3dab47 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_normalize_winnow/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_normalize_winnow/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_normalize_winnow/RandomForest/training_metadata.json new file mode 100644 index 00000000..e86cf50a --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_normalize_winnow/RandomForest/training_metadata.json @@ -0,0 +1,31 @@ +{ + "preprocessing_steps": [ + "eliminate", + "normalize", + "winnow" + ], + "mode": "both", + "cv_score": 0.475, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_normalize_winnow/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_normalize_winnow/SVM/model.joblib new file mode 100644 index 00000000..ecf33b57 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_normalize_winnow/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_normalize_winnow/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_normalize_winnow/SVM/training_metadata.json new file mode 100644 index 00000000..77819b24 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_normalize_winnow/SVM/training_metadata.json @@ -0,0 +1,27 @@ +{ + "preprocessing_steps": [ + "eliminate", + "normalize", + "winnow" + ], + "mode": "both", + "cv_score": 0.475, + "model_type": "SVC", + "hyperparameters": { + "C": 1, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_normalize_winnow/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_normalize_winnow/selected_features.json new file mode 100644 index 00000000..ae9e8e91 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_normalize_winnow/selected_features.json @@ -0,0 +1,66 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "logRelF0-H1-H2_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "mfcc1V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "VoicedSegmentsPerSec", + "StddevVoicedSegmentLengthSec", + "StddevUnvoicedSegmentLength", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_winnow/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_winnow/RandomForest/model.joblib new file mode 100644 index 00000000..8a6712db Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_winnow/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_winnow/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_winnow/RandomForest/training_metadata.json new file mode 100644 index 00000000..093b5bc8 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_winnow/RandomForest/training_metadata.json @@ -0,0 +1,30 @@ +{ + "preprocessing_steps": [ + "eliminate", + "winnow" + ], + "mode": null, + "cv_score": 0.325, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_winnow/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_winnow/SVM/model.joblib new file mode 100644 index 00000000..425d0647 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_winnow/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_winnow/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_winnow/SVM/training_metadata.json new file mode 100644 index 00000000..76e188b4 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_winnow/SVM/training_metadata.json @@ -0,0 +1,26 @@ +{ + "preprocessing_steps": [ + "eliminate", + "winnow" + ], + "mode": null, + "cv_score": 0.475, + "model_type": "SVC", + "hyperparameters": { + "C": 1, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_winnow/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_winnow/selected_features.json new file mode 100644 index 00000000..ae9e8e91 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_winnow/selected_features.json @@ -0,0 +1,66 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "logRelF0-H1-H2_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "mfcc1V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "VoicedSegmentsPerSec", + "StddevVoicedSegmentLengthSec", + "StddevUnvoicedSegmentLength", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_winnow_normalize/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_winnow_normalize/RandomForest/model.joblib new file mode 100644 index 00000000..5f820b73 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_winnow_normalize/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_winnow_normalize/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_winnow_normalize/RandomForest/training_metadata.json new file mode 100644 index 00000000..cd33ce38 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_winnow_normalize/RandomForest/training_metadata.json @@ -0,0 +1,31 @@ +{ + "preprocessing_steps": [ + "eliminate", + "winnow", + "normalize" + ], + "mode": "both", + "cv_score": 0.425, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_winnow_normalize/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_winnow_normalize/SVM/model.joblib new file mode 100644 index 00000000..df5f2652 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_winnow_normalize/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_winnow_normalize/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_winnow_normalize/SVM/training_metadata.json new file mode 100644 index 00000000..f3597191 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_winnow_normalize/SVM/training_metadata.json @@ -0,0 +1,27 @@ +{ + "preprocessing_steps": [ + "eliminate", + "winnow", + "normalize" + ], + "mode": "both", + "cv_score": 0.475, + "model_type": "SVC", + "hyperparameters": { + "C": 1, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_winnow_normalize/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_winnow_normalize/selected_features.json new file mode 100644 index 00000000..ae9e8e91 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/eliminate_winnow_normalize/selected_features.json @@ -0,0 +1,66 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "logRelF0-H1-H2_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "mfcc1V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "VoicedSegmentsPerSec", + "StddevVoicedSegmentLengthSec", + "StddevUnvoicedSegmentLength", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize/RandomForest/model.joblib new file mode 100644 index 00000000..e3733033 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize/RandomForest/training_metadata.json new file mode 100644 index 00000000..7e3a39c1 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize/RandomForest/training_metadata.json @@ -0,0 +1,29 @@ +{ + "preprocessing_steps": [ + "normalize" + ], + "mode": "both", + "cv_score": 0.325, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize/SVM/model.joblib new file mode 100644 index 00000000..0a169dd0 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize/SVM/training_metadata.json new file mode 100644 index 00000000..6d5455ee --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize/SVM/training_metadata.json @@ -0,0 +1,25 @@ +{ + "preprocessing_steps": [ + "normalize" + ], + "mode": "both", + "cv_score": 0.325, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_eliminate/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_eliminate/RandomForest/model.joblib new file mode 100644 index 00000000..1bfa9e87 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_eliminate/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_eliminate/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_eliminate/RandomForest/training_metadata.json new file mode 100644 index 00000000..26b37d16 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_eliminate/RandomForest/training_metadata.json @@ -0,0 +1,30 @@ +{ + "preprocessing_steps": [ + "normalize", + "eliminate" + ], + "mode": "both", + "cv_score": 0.275, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_eliminate/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_eliminate/SVM/model.joblib new file mode 100644 index 00000000..01f84d3d Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_eliminate/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_eliminate/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_eliminate/SVM/training_metadata.json new file mode 100644 index 00000000..55323697 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_eliminate/SVM/training_metadata.json @@ -0,0 +1,26 @@ +{ + "preprocessing_steps": [ + "normalize", + "eliminate" + ], + "mode": "both", + "cv_score": 0.45, + "model_type": "SVC", + "hyperparameters": { + "C": 1, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_eliminate/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_eliminate/selected_features.json new file mode 100644 index 00000000..ae9e8e91 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_eliminate/selected_features.json @@ -0,0 +1,66 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "logRelF0-H1-H2_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "mfcc1V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "VoicedSegmentsPerSec", + "StddevVoicedSegmentLengthSec", + "StddevUnvoicedSegmentLength", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_eliminate_winnow/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_eliminate_winnow/RandomForest/model.joblib new file mode 100644 index 00000000..a099a7ba Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_eliminate_winnow/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_eliminate_winnow/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_eliminate_winnow/RandomForest/training_metadata.json new file mode 100644 index 00000000..4b93f0a4 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_eliminate_winnow/RandomForest/training_metadata.json @@ -0,0 +1,31 @@ +{ + "preprocessing_steps": [ + "normalize", + "eliminate", + "winnow" + ], + "mode": "both", + "cv_score": 0.325, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_eliminate_winnow/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_eliminate_winnow/SVM/model.joblib new file mode 100644 index 00000000..ea121e07 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_eliminate_winnow/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_eliminate_winnow/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_eliminate_winnow/SVM/training_metadata.json new file mode 100644 index 00000000..90f22583 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_eliminate_winnow/SVM/training_metadata.json @@ -0,0 +1,27 @@ +{ + "preprocessing_steps": [ + "normalize", + "eliminate", + "winnow" + ], + "mode": "both", + "cv_score": 0.45, + "model_type": "SVC", + "hyperparameters": { + "C": 1, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_eliminate_winnow/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_eliminate_winnow/selected_features.json new file mode 100644 index 00000000..ae9e8e91 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_eliminate_winnow/selected_features.json @@ -0,0 +1,66 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "logRelF0-H1-H2_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "mfcc1V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "VoicedSegmentsPerSec", + "StddevVoicedSegmentLengthSec", + "StddevUnvoicedSegmentLength", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_winnow/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_winnow/RandomForest/model.joblib new file mode 100644 index 00000000..1b366268 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_winnow/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_winnow/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_winnow/RandomForest/training_metadata.json new file mode 100644 index 00000000..9c0ec5fc --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_winnow/RandomForest/training_metadata.json @@ -0,0 +1,30 @@ +{ + "preprocessing_steps": [ + "normalize", + "winnow" + ], + "mode": "both", + "cv_score": 0.125, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_winnow/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_winnow/SVM/model.joblib new file mode 100644 index 00000000..f9500483 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_winnow/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_winnow/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_winnow/SVM/training_metadata.json new file mode 100644 index 00000000..ec2642be --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_winnow/SVM/training_metadata.json @@ -0,0 +1,26 @@ +{ + "preprocessing_steps": [ + "normalize", + "winnow" + ], + "mode": "both", + "cv_score": 0.35, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_winnow/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_winnow/selected_features.json new file mode 100644 index 00000000..ae9e8e91 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_winnow/selected_features.json @@ -0,0 +1,66 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "logRelF0-H1-H2_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "mfcc1V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "VoicedSegmentsPerSec", + "StddevVoicedSegmentLengthSec", + "StddevUnvoicedSegmentLength", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_winnow_eliminate/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_winnow_eliminate/RandomForest/model.joblib new file mode 100644 index 00000000..1b366268 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_winnow_eliminate/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_winnow_eliminate/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_winnow_eliminate/RandomForest/training_metadata.json new file mode 100644 index 00000000..7e1945d5 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_winnow_eliminate/RandomForest/training_metadata.json @@ -0,0 +1,31 @@ +{ + "preprocessing_steps": [ + "normalize", + "winnow", + "eliminate" + ], + "mode": "both", + "cv_score": 0.125, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_winnow_eliminate/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_winnow_eliminate/SVM/model.joblib new file mode 100644 index 00000000..f9500483 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_winnow_eliminate/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_winnow_eliminate/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_winnow_eliminate/SVM/training_metadata.json new file mode 100644 index 00000000..b05bc8a2 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_winnow_eliminate/SVM/training_metadata.json @@ -0,0 +1,27 @@ +{ + "preprocessing_steps": [ + "normalize", + "winnow", + "eliminate" + ], + "mode": "both", + "cv_score": 0.35, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_winnow_eliminate/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_winnow_eliminate/selected_features.json new file mode 100644 index 00000000..ae9e8e91 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/normalize_winnow_eliminate/selected_features.json @@ -0,0 +1,66 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "logRelF0-H1-H2_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "mfcc1V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "VoicedSegmentsPerSec", + "StddevVoicedSegmentLengthSec", + "StddevUnvoicedSegmentLength", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow/RandomForest/model.joblib new file mode 100644 index 00000000..51c2b511 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow/RandomForest/training_metadata.json new file mode 100644 index 00000000..453df33e --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow/RandomForest/training_metadata.json @@ -0,0 +1,29 @@ +{ + "preprocessing_steps": [ + "winnow" + ], + "mode": null, + "cv_score": 0.4, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow/SVM/model.joblib new file mode 100644 index 00000000..799bdc1a Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow/SVM/training_metadata.json new file mode 100644 index 00000000..178327b7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow/SVM/training_metadata.json @@ -0,0 +1,25 @@ +{ + "preprocessing_steps": [ + "winnow" + ], + "mode": null, + "cv_score": 0.4, + "model_type": "SVC", + "hyperparameters": { + "C": 1, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow/selected_features.json new file mode 100644 index 00000000..7858a341 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow/selected_features.json @@ -0,0 +1,66 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "shimmerLocaldB_sma3nz_amean", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_stddevNorm", + "mfcc1V_sma3nz_stddevNorm", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "StddevUnvoicedSegmentLength", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_eliminate/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_eliminate/RandomForest/model.joblib new file mode 100644 index 00000000..51c2b511 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_eliminate/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_eliminate/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_eliminate/RandomForest/training_metadata.json new file mode 100644 index 00000000..3e14b253 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_eliminate/RandomForest/training_metadata.json @@ -0,0 +1,30 @@ +{ + "preprocessing_steps": [ + "winnow", + "eliminate" + ], + "mode": null, + "cv_score": 0.4, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_eliminate/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_eliminate/SVM/model.joblib new file mode 100644 index 00000000..799bdc1a Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_eliminate/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_eliminate/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_eliminate/SVM/training_metadata.json new file mode 100644 index 00000000..f778bd02 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_eliminate/SVM/training_metadata.json @@ -0,0 +1,26 @@ +{ + "preprocessing_steps": [ + "winnow", + "eliminate" + ], + "mode": null, + "cv_score": 0.4, + "model_type": "SVC", + "hyperparameters": { + "C": 1, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_eliminate/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_eliminate/selected_features.json new file mode 100644 index 00000000..ae9e8e91 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_eliminate/selected_features.json @@ -0,0 +1,66 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "logRelF0-H1-H2_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "mfcc1V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "VoicedSegmentsPerSec", + "StddevVoicedSegmentLengthSec", + "StddevUnvoicedSegmentLength", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_eliminate_normalize/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_eliminate_normalize/RandomForest/model.joblib new file mode 100644 index 00000000..bbf4e22a Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_eliminate_normalize/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_eliminate_normalize/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_eliminate_normalize/RandomForest/training_metadata.json new file mode 100644 index 00000000..79bfe9ea --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_eliminate_normalize/RandomForest/training_metadata.json @@ -0,0 +1,31 @@ +{ + "preprocessing_steps": [ + "winnow", + "eliminate", + "normalize" + ], + "mode": "both", + "cv_score": 0.15, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_eliminate_normalize/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_eliminate_normalize/SVM/model.joblib new file mode 100644 index 00000000..9674459f Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_eliminate_normalize/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_eliminate_normalize/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_eliminate_normalize/SVM/training_metadata.json new file mode 100644 index 00000000..d57a7d6e --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_eliminate_normalize/SVM/training_metadata.json @@ -0,0 +1,27 @@ +{ + "preprocessing_steps": [ + "winnow", + "eliminate", + "normalize" + ], + "mode": "both", + "cv_score": 0.35, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_eliminate_normalize/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_eliminate_normalize/selected_features.json new file mode 100644 index 00000000..ae9e8e91 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_eliminate_normalize/selected_features.json @@ -0,0 +1,66 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "logRelF0-H1-H2_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "mfcc1V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "VoicedSegmentsPerSec", + "StddevVoicedSegmentLengthSec", + "StddevUnvoicedSegmentLength", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_normalize/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_normalize/RandomForest/model.joblib new file mode 100644 index 00000000..abf640c4 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_normalize/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_normalize/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_normalize/RandomForest/training_metadata.json new file mode 100644 index 00000000..b51b36a7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_normalize/RandomForest/training_metadata.json @@ -0,0 +1,30 @@ +{ + "preprocessing_steps": [ + "winnow", + "normalize" + ], + "mode": "both", + "cv_score": 0.2, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_normalize/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_normalize/SVM/model.joblib new file mode 100644 index 00000000..8394741b Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_normalize/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_normalize/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_normalize/SVM/training_metadata.json new file mode 100644 index 00000000..31a5a5ba --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_normalize/SVM/training_metadata.json @@ -0,0 +1,26 @@ +{ + "preprocessing_steps": [ + "winnow", + "normalize" + ], + "mode": "both", + "cv_score": 0.35, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_normalize/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_normalize/selected_features.json new file mode 100644 index 00000000..ae9e8e91 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_normalize/selected_features.json @@ -0,0 +1,66 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "logRelF0-H1-H2_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "mfcc1V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "VoicedSegmentsPerSec", + "StddevVoicedSegmentLengthSec", + "StddevUnvoicedSegmentLength", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_normalize_eliminate/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_normalize_eliminate/RandomForest/model.joblib new file mode 100644 index 00000000..abf640c4 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_normalize_eliminate/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_normalize_eliminate/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_normalize_eliminate/RandomForest/training_metadata.json new file mode 100644 index 00000000..1641024b --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_normalize_eliminate/RandomForest/training_metadata.json @@ -0,0 +1,31 @@ +{ + "preprocessing_steps": [ + "winnow", + "normalize", + "eliminate" + ], + "mode": "both", + "cv_score": 0.2, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_normalize_eliminate/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_normalize_eliminate/SVM/model.joblib new file mode 100644 index 00000000..8394741b Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_normalize_eliminate/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_normalize_eliminate/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_normalize_eliminate/SVM/training_metadata.json new file mode 100644 index 00000000..48e58570 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_normalize_eliminate/SVM/training_metadata.json @@ -0,0 +1,27 @@ +{ + "preprocessing_steps": [ + "winnow", + "normalize", + "eliminate" + ], + "mode": "both", + "cv_score": 0.35, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_normalize_eliminate/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_normalize_eliminate/selected_features.json new file mode 100644 index 00000000..ae9e8e91 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/inner_loop/winnow_normalize_eliminate/selected_features.json @@ -0,0 +1,66 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "logRelF0-H1-H2_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "mfcc1V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "VoicedSegmentsPerSec", + "StddevVoicedSegmentLengthSec", + "StddevUnvoicedSegmentLength", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/model.joblib new file mode 100644 index 00000000..135597cd Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/training_metadata.json new file mode 100644 index 00000000..2be2b392 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_MIT/training_metadata.json @@ -0,0 +1,95 @@ +{ + "site": "MIT", + "best_preprocessing_steps": [ + [ + "normalize", + "eliminate" + ], + "scale" + ], + "test_score": 0.3, + "model_type": "SVC", + "hyperparameters": { + "C": 1, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + }, + "selected_features": [ + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "logRelF0-H1-H2_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "mfcc1V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "VoicedSegmentsPerSec", + "StddevVoicedSegmentLengthSec", + "StddevUnvoicedSegmentLength", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" + ] +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate/RandomForest/model.joblib new file mode 100644 index 00000000..a24ad96b Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate/RandomForest/training_metadata.json new file mode 100644 index 00000000..df74c96d --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate/RandomForest/training_metadata.json @@ -0,0 +1,29 @@ +{ + "preprocessing_steps": [ + "eliminate" + ], + "mode": null, + "cv_score": 0.375, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate/SVM/model.joblib new file mode 100644 index 00000000..1d6ebce9 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate/SVM/training_metadata.json new file mode 100644 index 00000000..3ef560c5 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate/SVM/training_metadata.json @@ -0,0 +1,25 @@ +{ + "preprocessing_steps": [ + "eliminate" + ], + "mode": null, + "cv_score": 0.425, + "model_type": "SVC", + "hyperparameters": { + "C": 1, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_normalize/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_normalize/RandomForest/model.joblib new file mode 100644 index 00000000..b86515cf Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_normalize/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_normalize/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_normalize/RandomForest/training_metadata.json new file mode 100644 index 00000000..faed1e39 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_normalize/RandomForest/training_metadata.json @@ -0,0 +1,30 @@ +{ + "preprocessing_steps": [ + "eliminate", + "normalize" + ], + "mode": "both", + "cv_score": 0.425, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_normalize/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_normalize/SVM/model.joblib new file mode 100644 index 00000000..bd28a54f Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_normalize/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_normalize/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_normalize/SVM/training_metadata.json new file mode 100644 index 00000000..3285a8b5 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_normalize/SVM/training_metadata.json @@ -0,0 +1,26 @@ +{ + "preprocessing_steps": [ + "eliminate", + "normalize" + ], + "mode": "both", + "cv_score": 0.475, + "model_type": "SVC", + "hyperparameters": { + "C": 1, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_normalize/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_normalize/selected_features.json new file mode 100644 index 00000000..ee5a5961 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_normalize/selected_features.json @@ -0,0 +1,66 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "HNRdBACF_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_stddevNorm", + "mfcc1V_sma3nz_stddevNorm", + "MeanVoicedSegmentLengthSec", + "StddevUnvoicedSegmentLength", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_normalize_winnow/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_normalize_winnow/RandomForest/model.joblib new file mode 100644 index 00000000..d188afc7 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_normalize_winnow/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_normalize_winnow/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_normalize_winnow/RandomForest/training_metadata.json new file mode 100644 index 00000000..c9011498 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_normalize_winnow/RandomForest/training_metadata.json @@ -0,0 +1,31 @@ +{ + "preprocessing_steps": [ + "eliminate", + "normalize", + "winnow" + ], + "mode": "both", + "cv_score": 0.4, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_normalize_winnow/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_normalize_winnow/SVM/model.joblib new file mode 100644 index 00000000..08b35939 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_normalize_winnow/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_normalize_winnow/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_normalize_winnow/SVM/training_metadata.json new file mode 100644 index 00000000..77819b24 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_normalize_winnow/SVM/training_metadata.json @@ -0,0 +1,27 @@ +{ + "preprocessing_steps": [ + "eliminate", + "normalize", + "winnow" + ], + "mode": "both", + "cv_score": 0.475, + "model_type": "SVC", + "hyperparameters": { + "C": 1, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_normalize_winnow/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_normalize_winnow/selected_features.json new file mode 100644 index 00000000..ee5a5961 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_normalize_winnow/selected_features.json @@ -0,0 +1,66 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "HNRdBACF_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_stddevNorm", + "mfcc1V_sma3nz_stddevNorm", + "MeanVoicedSegmentLengthSec", + "StddevUnvoicedSegmentLength", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_winnow/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_winnow/RandomForest/model.joblib new file mode 100644 index 00000000..6c04b9d2 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_winnow/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_winnow/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_winnow/RandomForest/training_metadata.json new file mode 100644 index 00000000..c21939c1 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_winnow/RandomForest/training_metadata.json @@ -0,0 +1,30 @@ +{ + "preprocessing_steps": [ + "eliminate", + "winnow" + ], + "mode": null, + "cv_score": 0.175, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_winnow/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_winnow/SVM/model.joblib new file mode 100644 index 00000000..5502ad74 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_winnow/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_winnow/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_winnow/SVM/training_metadata.json new file mode 100644 index 00000000..9e6d6a25 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_winnow/SVM/training_metadata.json @@ -0,0 +1,26 @@ +{ + "preprocessing_steps": [ + "eliminate", + "winnow" + ], + "mode": null, + "cv_score": 0.425, + "model_type": "SVC", + "hyperparameters": { + "C": 1, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_winnow/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_winnow/selected_features.json new file mode 100644 index 00000000..ee5a5961 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_winnow/selected_features.json @@ -0,0 +1,66 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "HNRdBACF_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_stddevNorm", + "mfcc1V_sma3nz_stddevNorm", + "MeanVoicedSegmentLengthSec", + "StddevUnvoicedSegmentLength", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_winnow_normalize/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_winnow_normalize/RandomForest/model.joblib new file mode 100644 index 00000000..2c48a3a2 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_winnow_normalize/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_winnow_normalize/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_winnow_normalize/RandomForest/training_metadata.json new file mode 100644 index 00000000..7e4deca1 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_winnow_normalize/RandomForest/training_metadata.json @@ -0,0 +1,31 @@ +{ + "preprocessing_steps": [ + "eliminate", + "winnow", + "normalize" + ], + "mode": "both", + "cv_score": 0.375, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_winnow_normalize/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_winnow_normalize/SVM/model.joblib new file mode 100644 index 00000000..19deed77 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_winnow_normalize/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_winnow_normalize/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_winnow_normalize/SVM/training_metadata.json new file mode 100644 index 00000000..f3597191 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_winnow_normalize/SVM/training_metadata.json @@ -0,0 +1,27 @@ +{ + "preprocessing_steps": [ + "eliminate", + "winnow", + "normalize" + ], + "mode": "both", + "cv_score": 0.475, + "model_type": "SVC", + "hyperparameters": { + "C": 1, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_winnow_normalize/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_winnow_normalize/selected_features.json new file mode 100644 index 00000000..ee5a5961 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/eliminate_winnow_normalize/selected_features.json @@ -0,0 +1,66 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "HNRdBACF_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_stddevNorm", + "mfcc1V_sma3nz_stddevNorm", + "MeanVoicedSegmentLengthSec", + "StddevUnvoicedSegmentLength", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize/RandomForest/model.joblib new file mode 100644 index 00000000..1a1f56d9 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize/RandomForest/training_metadata.json new file mode 100644 index 00000000..0a5bf3e8 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize/RandomForest/training_metadata.json @@ -0,0 +1,29 @@ +{ + "preprocessing_steps": [ + "normalize" + ], + "mode": "both", + "cv_score": 0.25, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize/SVM/model.joblib new file mode 100644 index 00000000..d2dbd51b Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize/SVM/training_metadata.json new file mode 100644 index 00000000..e2d589a9 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize/SVM/training_metadata.json @@ -0,0 +1,25 @@ +{ + "preprocessing_steps": [ + "normalize" + ], + "mode": "both", + "cv_score": 0.425, + "model_type": "SVC", + "hyperparameters": { + "C": 1, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_eliminate/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_eliminate/RandomForest/model.joblib new file mode 100644 index 00000000..0a9348e4 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_eliminate/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_eliminate/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_eliminate/RandomForest/training_metadata.json new file mode 100644 index 00000000..3951cbf8 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_eliminate/RandomForest/training_metadata.json @@ -0,0 +1,30 @@ +{ + "preprocessing_steps": [ + "normalize", + "eliminate" + ], + "mode": "both", + "cv_score": 0.35, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_eliminate/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_eliminate/SVM/model.joblib new file mode 100644 index 00000000..2817c014 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_eliminate/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_eliminate/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_eliminate/SVM/training_metadata.json new file mode 100644 index 00000000..3fd49d23 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_eliminate/SVM/training_metadata.json @@ -0,0 +1,26 @@ +{ + "preprocessing_steps": [ + "normalize", + "eliminate" + ], + "mode": "both", + "cv_score": 0.375, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_eliminate/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_eliminate/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_eliminate/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_eliminate_winnow/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_eliminate_winnow/RandomForest/model.joblib new file mode 100644 index 00000000..94902a37 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_eliminate_winnow/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_eliminate_winnow/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_eliminate_winnow/RandomForest/training_metadata.json new file mode 100644 index 00000000..001d9a3a --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_eliminate_winnow/RandomForest/training_metadata.json @@ -0,0 +1,31 @@ +{ + "preprocessing_steps": [ + "normalize", + "eliminate", + "winnow" + ], + "mode": "both", + "cv_score": 0.375, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_eliminate_winnow/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_eliminate_winnow/SVM/model.joblib new file mode 100644 index 00000000..edeefa70 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_eliminate_winnow/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_eliminate_winnow/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_eliminate_winnow/SVM/training_metadata.json new file mode 100644 index 00000000..f8f6be17 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_eliminate_winnow/SVM/training_metadata.json @@ -0,0 +1,27 @@ +{ + "preprocessing_steps": [ + "normalize", + "eliminate", + "winnow" + ], + "mode": "both", + "cv_score": 0.375, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_eliminate_winnow/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_eliminate_winnow/selected_features.json new file mode 100644 index 00000000..ee5a5961 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_eliminate_winnow/selected_features.json @@ -0,0 +1,66 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "HNRdBACF_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_stddevNorm", + "mfcc1V_sma3nz_stddevNorm", + "MeanVoicedSegmentLengthSec", + "StddevUnvoicedSegmentLength", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_winnow/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_winnow/RandomForest/model.joblib new file mode 100644 index 00000000..6e94775e Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_winnow/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_winnow/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_winnow/RandomForest/training_metadata.json new file mode 100644 index 00000000..4196a43b --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_winnow/RandomForest/training_metadata.json @@ -0,0 +1,30 @@ +{ + "preprocessing_steps": [ + "normalize", + "winnow" + ], + "mode": "both", + "cv_score": 0.425, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_winnow/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_winnow/SVM/model.joblib new file mode 100644 index 00000000..dccec0e7 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_winnow/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_winnow/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_winnow/SVM/training_metadata.json new file mode 100644 index 00000000..fa5c42a5 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_winnow/SVM/training_metadata.json @@ -0,0 +1,26 @@ +{ + "preprocessing_steps": [ + "normalize", + "winnow" + ], + "mode": "both", + "cv_score": 0.375, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_winnow/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_winnow/selected_features.json new file mode 100644 index 00000000..ee5a5961 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_winnow/selected_features.json @@ -0,0 +1,66 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "HNRdBACF_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_stddevNorm", + "mfcc1V_sma3nz_stddevNorm", + "MeanVoicedSegmentLengthSec", + "StddevUnvoicedSegmentLength", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_winnow_eliminate/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_winnow_eliminate/RandomForest/model.joblib new file mode 100644 index 00000000..6e94775e Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_winnow_eliminate/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_winnow_eliminate/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_winnow_eliminate/RandomForest/training_metadata.json new file mode 100644 index 00000000..a29b09d7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_winnow_eliminate/RandomForest/training_metadata.json @@ -0,0 +1,31 @@ +{ + "preprocessing_steps": [ + "normalize", + "winnow", + "eliminate" + ], + "mode": "both", + "cv_score": 0.425, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_winnow_eliminate/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_winnow_eliminate/SVM/model.joblib new file mode 100644 index 00000000..dccec0e7 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_winnow_eliminate/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_winnow_eliminate/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_winnow_eliminate/SVM/training_metadata.json new file mode 100644 index 00000000..5998c2f4 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_winnow_eliminate/SVM/training_metadata.json @@ -0,0 +1,27 @@ +{ + "preprocessing_steps": [ + "normalize", + "winnow", + "eliminate" + ], + "mode": "both", + "cv_score": 0.375, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_winnow_eliminate/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_winnow_eliminate/selected_features.json new file mode 100644 index 00000000..ee5a5961 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/normalize_winnow_eliminate/selected_features.json @@ -0,0 +1,66 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "HNRdBACF_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_stddevNorm", + "mfcc1V_sma3nz_stddevNorm", + "MeanVoicedSegmentLengthSec", + "StddevUnvoicedSegmentLength", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow/RandomForest/model.joblib new file mode 100644 index 00000000..dc96332f Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow/RandomForest/training_metadata.json new file mode 100644 index 00000000..b0acfd48 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow/RandomForest/training_metadata.json @@ -0,0 +1,29 @@ +{ + "preprocessing_steps": [ + "winnow" + ], + "mode": null, + "cv_score": 0.35, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow/SVM/model.joblib new file mode 100644 index 00000000..e42b2615 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow/SVM/training_metadata.json new file mode 100644 index 00000000..06619ec5 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow/SVM/training_metadata.json @@ -0,0 +1,25 @@ +{ + "preprocessing_steps": [ + "winnow" + ], + "mode": null, + "cv_score": 0.375, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_eliminate/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_eliminate/RandomForest/model.joblib new file mode 100644 index 00000000..dc96332f Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_eliminate/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_eliminate/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_eliminate/RandomForest/training_metadata.json new file mode 100644 index 00000000..eb159e49 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_eliminate/RandomForest/training_metadata.json @@ -0,0 +1,30 @@ +{ + "preprocessing_steps": [ + "winnow", + "eliminate" + ], + "mode": null, + "cv_score": 0.35, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_eliminate/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_eliminate/SVM/model.joblib new file mode 100644 index 00000000..e42b2615 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_eliminate/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_eliminate/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_eliminate/SVM/training_metadata.json new file mode 100644 index 00000000..59714289 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_eliminate/SVM/training_metadata.json @@ -0,0 +1,26 @@ +{ + "preprocessing_steps": [ + "winnow", + "eliminate" + ], + "mode": null, + "cv_score": 0.375, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_eliminate/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_eliminate/selected_features.json new file mode 100644 index 00000000..ee5a5961 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_eliminate/selected_features.json @@ -0,0 +1,66 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "HNRdBACF_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_stddevNorm", + "mfcc1V_sma3nz_stddevNorm", + "MeanVoicedSegmentLengthSec", + "StddevUnvoicedSegmentLength", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_eliminate_normalize/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_eliminate_normalize/RandomForest/model.joblib new file mode 100644 index 00000000..87e9e665 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_eliminate_normalize/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_eliminate_normalize/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_eliminate_normalize/RandomForest/training_metadata.json new file mode 100644 index 00000000..2cb2d432 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_eliminate_normalize/RandomForest/training_metadata.json @@ -0,0 +1,31 @@ +{ + "preprocessing_steps": [ + "winnow", + "eliminate", + "normalize" + ], + "mode": "both", + "cv_score": 0.325, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_eliminate_normalize/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_eliminate_normalize/SVM/model.joblib new file mode 100644 index 00000000..090f9a2b Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_eliminate_normalize/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_eliminate_normalize/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_eliminate_normalize/SVM/training_metadata.json new file mode 100644 index 00000000..8958c946 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_eliminate_normalize/SVM/training_metadata.json @@ -0,0 +1,27 @@ +{ + "preprocessing_steps": [ + "winnow", + "eliminate", + "normalize" + ], + "mode": "both", + "cv_score": 0.375, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_eliminate_normalize/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_eliminate_normalize/selected_features.json new file mode 100644 index 00000000..ee5a5961 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_eliminate_normalize/selected_features.json @@ -0,0 +1,66 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "HNRdBACF_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_stddevNorm", + "mfcc1V_sma3nz_stddevNorm", + "MeanVoicedSegmentLengthSec", + "StddevUnvoicedSegmentLength", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_normalize/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_normalize/RandomForest/model.joblib new file mode 100644 index 00000000..ee5f749d Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_normalize/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_normalize/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_normalize/RandomForest/training_metadata.json new file mode 100644 index 00000000..267e47f4 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_normalize/RandomForest/training_metadata.json @@ -0,0 +1,30 @@ +{ + "preprocessing_steps": [ + "winnow", + "normalize" + ], + "mode": "both", + "cv_score": 0.325, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_normalize/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_normalize/SVM/model.joblib new file mode 100644 index 00000000..6f3ce056 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_normalize/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_normalize/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_normalize/SVM/training_metadata.json new file mode 100644 index 00000000..d9930d89 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_normalize/SVM/training_metadata.json @@ -0,0 +1,26 @@ +{ + "preprocessing_steps": [ + "winnow", + "normalize" + ], + "mode": "both", + "cv_score": 0.375, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_normalize/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_normalize/selected_features.json new file mode 100644 index 00000000..ee5a5961 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_normalize/selected_features.json @@ -0,0 +1,66 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "HNRdBACF_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_stddevNorm", + "mfcc1V_sma3nz_stddevNorm", + "MeanVoicedSegmentLengthSec", + "StddevUnvoicedSegmentLength", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_normalize_eliminate/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_normalize_eliminate/RandomForest/model.joblib new file mode 100644 index 00000000..ee5f749d Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_normalize_eliminate/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_normalize_eliminate/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_normalize_eliminate/RandomForest/training_metadata.json new file mode 100644 index 00000000..d0cd5505 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_normalize_eliminate/RandomForest/training_metadata.json @@ -0,0 +1,31 @@ +{ + "preprocessing_steps": [ + "winnow", + "normalize", + "eliminate" + ], + "mode": "both", + "cv_score": 0.325, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_normalize_eliminate/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_normalize_eliminate/SVM/model.joblib new file mode 100644 index 00000000..6f3ce056 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_normalize_eliminate/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_normalize_eliminate/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_normalize_eliminate/SVM/training_metadata.json new file mode 100644 index 00000000..35a0d394 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_normalize_eliminate/SVM/training_metadata.json @@ -0,0 +1,27 @@ +{ + "preprocessing_steps": [ + "winnow", + "normalize", + "eliminate" + ], + "mode": "both", + "cv_score": 0.375, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_normalize_eliminate/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_normalize_eliminate/selected_features.json new file mode 100644 index 00000000..ee5a5961 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/inner_loop/winnow_normalize_eliminate/selected_features.json @@ -0,0 +1,66 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "HNRdBACF_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_stddevNorm", + "mfcc1V_sma3nz_stddevNorm", + "MeanVoicedSegmentLengthSec", + "StddevUnvoicedSegmentLength", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/model.joblib new file mode 100644 index 00000000..d568c832 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/training_metadata.json new file mode 100644 index 00000000..5a90ae06 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_Mt. Sinai/training_metadata.json @@ -0,0 +1,95 @@ +{ + "site": "Mt. Sinai", + "best_preprocessing_steps": [ + [ + "eliminate", + "normalize" + ], + "scale" + ], + "test_score": 0.4, + "model_type": "SVC", + "hyperparameters": { + "C": 1, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + }, + "selected_features": [ + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "HNRdBACF_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_stddevNorm", + "mfcc1V_sma3nz_stddevNorm", + "MeanVoicedSegmentLengthSec", + "StddevUnvoicedSegmentLength", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" + ] +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate/RandomForest/model.joblib new file mode 100644 index 00000000..5f2a1a19 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate/RandomForest/training_metadata.json new file mode 100644 index 00000000..f1d1b661 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate/RandomForest/training_metadata.json @@ -0,0 +1,29 @@ +{ + "preprocessing_steps": [ + "eliminate" + ], + "mode": null, + "cv_score": 0.325, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate/SVM/model.joblib new file mode 100644 index 00000000..c398d9cd Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate/SVM/training_metadata.json new file mode 100644 index 00000000..789cd2ae --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate/SVM/training_metadata.json @@ -0,0 +1,25 @@ +{ + "preprocessing_steps": [ + "eliminate" + ], + "mode": null, + "cv_score": 0.4, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_normalize/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_normalize/RandomForest/model.joblib new file mode 100644 index 00000000..7fb0c56b Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_normalize/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_normalize/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_normalize/RandomForest/training_metadata.json new file mode 100644 index 00000000..65095208 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_normalize/RandomForest/training_metadata.json @@ -0,0 +1,30 @@ +{ + "preprocessing_steps": [ + "eliminate", + "normalize" + ], + "mode": "both", + "cv_score": 0.4, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_normalize/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_normalize/SVM/model.joblib new file mode 100644 index 00000000..078abab2 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_normalize/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_normalize/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_normalize/SVM/training_metadata.json new file mode 100644 index 00000000..23fd2ca4 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_normalize/SVM/training_metadata.json @@ -0,0 +1,26 @@ +{ + "preprocessing_steps": [ + "eliminate", + "normalize" + ], + "mode": "both", + "cv_score": 0.4, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_normalize/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_normalize/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_normalize/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_normalize_winnow/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_normalize_winnow/RandomForest/model.joblib new file mode 100644 index 00000000..55f54142 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_normalize_winnow/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_normalize_winnow/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_normalize_winnow/RandomForest/training_metadata.json new file mode 100644 index 00000000..c9011498 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_normalize_winnow/RandomForest/training_metadata.json @@ -0,0 +1,31 @@ +{ + "preprocessing_steps": [ + "eliminate", + "normalize", + "winnow" + ], + "mode": "both", + "cv_score": 0.4, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_normalize_winnow/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_normalize_winnow/SVM/model.joblib new file mode 100644 index 00000000..030360f1 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_normalize_winnow/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_normalize_winnow/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_normalize_winnow/SVM/training_metadata.json new file mode 100644 index 00000000..80ab93bf --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_normalize_winnow/SVM/training_metadata.json @@ -0,0 +1,27 @@ +{ + "preprocessing_steps": [ + "eliminate", + "normalize", + "winnow" + ], + "mode": "both", + "cv_score": 0.4, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_normalize_winnow/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_normalize_winnow/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_normalize_winnow/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_winnow/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_winnow/RandomForest/model.joblib new file mode 100644 index 00000000..01a9c747 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_winnow/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_winnow/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_winnow/RandomForest/training_metadata.json new file mode 100644 index 00000000..7b8e4a2a --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_winnow/RandomForest/training_metadata.json @@ -0,0 +1,30 @@ +{ + "preprocessing_steps": [ + "eliminate", + "winnow" + ], + "mode": null, + "cv_score": 0.3, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_winnow/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_winnow/SVM/model.joblib new file mode 100644 index 00000000..64db4db2 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_winnow/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_winnow/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_winnow/SVM/training_metadata.json new file mode 100644 index 00000000..e9c6b62a --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_winnow/SVM/training_metadata.json @@ -0,0 +1,26 @@ +{ + "preprocessing_steps": [ + "eliminate", + "winnow" + ], + "mode": null, + "cv_score": 0.4, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_winnow/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_winnow/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_winnow/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_winnow_normalize/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_winnow_normalize/RandomForest/model.joblib new file mode 100644 index 00000000..d98b91ea Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_winnow_normalize/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_winnow_normalize/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_winnow_normalize/RandomForest/training_metadata.json new file mode 100644 index 00000000..8da35e11 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_winnow_normalize/RandomForest/training_metadata.json @@ -0,0 +1,31 @@ +{ + "preprocessing_steps": [ + "eliminate", + "winnow", + "normalize" + ], + "mode": "both", + "cv_score": 0.475, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_winnow_normalize/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_winnow_normalize/SVM/model.joblib new file mode 100644 index 00000000..f788a76a Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_winnow_normalize/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_winnow_normalize/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_winnow_normalize/SVM/training_metadata.json new file mode 100644 index 00000000..9325d3f6 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_winnow_normalize/SVM/training_metadata.json @@ -0,0 +1,27 @@ +{ + "preprocessing_steps": [ + "eliminate", + "winnow", + "normalize" + ], + "mode": "both", + "cv_score": 0.4, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_winnow_normalize/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_winnow_normalize/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/eliminate_winnow_normalize/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize/RandomForest/model.joblib new file mode 100644 index 00000000..f37234af Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize/RandomForest/training_metadata.json new file mode 100644 index 00000000..0a5bf3e8 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize/RandomForest/training_metadata.json @@ -0,0 +1,29 @@ +{ + "preprocessing_steps": [ + "normalize" + ], + "mode": "both", + "cv_score": 0.25, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize/SVM/model.joblib new file mode 100644 index 00000000..3e4a1221 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize/SVM/training_metadata.json new file mode 100644 index 00000000..cee55117 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize/SVM/training_metadata.json @@ -0,0 +1,25 @@ +{ + "preprocessing_steps": [ + "normalize" + ], + "mode": "both", + "cv_score": 0.4, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_eliminate/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_eliminate/RandomForest/model.joblib new file mode 100644 index 00000000..008523b1 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_eliminate/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_eliminate/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_eliminate/RandomForest/training_metadata.json new file mode 100644 index 00000000..ed69eecf --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_eliminate/RandomForest/training_metadata.json @@ -0,0 +1,30 @@ +{ + "preprocessing_steps": [ + "normalize", + "eliminate" + ], + "mode": "both", + "cv_score": 0.425, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_eliminate/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_eliminate/SVM/model.joblib new file mode 100644 index 00000000..b37b1548 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_eliminate/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_eliminate/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_eliminate/SVM/training_metadata.json new file mode 100644 index 00000000..0d8762a4 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_eliminate/SVM/training_metadata.json @@ -0,0 +1,26 @@ +{ + "preprocessing_steps": [ + "normalize", + "eliminate" + ], + "mode": "both", + "cv_score": 0.4, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_eliminate/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_eliminate/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_eliminate/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_eliminate_winnow/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_eliminate_winnow/RandomForest/model.joblib new file mode 100644 index 00000000..43d60b29 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_eliminate_winnow/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_eliminate_winnow/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_eliminate_winnow/RandomForest/training_metadata.json new file mode 100644 index 00000000..4b93f0a4 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_eliminate_winnow/RandomForest/training_metadata.json @@ -0,0 +1,31 @@ +{ + "preprocessing_steps": [ + "normalize", + "eliminate", + "winnow" + ], + "mode": "both", + "cv_score": 0.325, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_eliminate_winnow/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_eliminate_winnow/SVM/model.joblib new file mode 100644 index 00000000..268818c1 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_eliminate_winnow/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_eliminate_winnow/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_eliminate_winnow/SVM/training_metadata.json new file mode 100644 index 00000000..082d6922 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_eliminate_winnow/SVM/training_metadata.json @@ -0,0 +1,27 @@ +{ + "preprocessing_steps": [ + "normalize", + "eliminate", + "winnow" + ], + "mode": "both", + "cv_score": 0.4, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_eliminate_winnow/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_eliminate_winnow/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_eliminate_winnow/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_winnow/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_winnow/RandomForest/model.joblib new file mode 100644 index 00000000..d34a9285 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_winnow/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_winnow/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_winnow/RandomForest/training_metadata.json new file mode 100644 index 00000000..53a27324 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_winnow/RandomForest/training_metadata.json @@ -0,0 +1,30 @@ +{ + "preprocessing_steps": [ + "normalize", + "winnow" + ], + "mode": "both", + "cv_score": 0.3, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_winnow/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_winnow/SVM/model.joblib new file mode 100644 index 00000000..2ad2431f Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_winnow/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_winnow/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_winnow/SVM/training_metadata.json new file mode 100644 index 00000000..9614e0a1 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_winnow/SVM/training_metadata.json @@ -0,0 +1,26 @@ +{ + "preprocessing_steps": [ + "normalize", + "winnow" + ], + "mode": "both", + "cv_score": 0.45, + "model_type": "SVC", + "hyperparameters": { + "C": 1, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_winnow/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_winnow/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_winnow/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_winnow_eliminate/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_winnow_eliminate/RandomForest/model.joblib new file mode 100644 index 00000000..d34a9285 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_winnow_eliminate/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_winnow_eliminate/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_winnow_eliminate/RandomForest/training_metadata.json new file mode 100644 index 00000000..1d1fe5be --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_winnow_eliminate/RandomForest/training_metadata.json @@ -0,0 +1,31 @@ +{ + "preprocessing_steps": [ + "normalize", + "winnow", + "eliminate" + ], + "mode": "both", + "cv_score": 0.3, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_winnow_eliminate/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_winnow_eliminate/SVM/model.joblib new file mode 100644 index 00000000..2ad2431f Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_winnow_eliminate/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_winnow_eliminate/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_winnow_eliminate/SVM/training_metadata.json new file mode 100644 index 00000000..28976565 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_winnow_eliminate/SVM/training_metadata.json @@ -0,0 +1,27 @@ +{ + "preprocessing_steps": [ + "normalize", + "winnow", + "eliminate" + ], + "mode": "both", + "cv_score": 0.45, + "model_type": "SVC", + "hyperparameters": { + "C": 1, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_winnow_eliminate/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_winnow_eliminate/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/normalize_winnow_eliminate/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow/RandomForest/model.joblib new file mode 100644 index 00000000..111b80bd Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow/RandomForest/training_metadata.json new file mode 100644 index 00000000..2273410f --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow/RandomForest/training_metadata.json @@ -0,0 +1,29 @@ +{ + "preprocessing_steps": [ + "winnow" + ], + "mode": null, + "cv_score": 0.375, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow/SVM/model.joblib new file mode 100644 index 00000000..c0998ade Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow/SVM/training_metadata.json new file mode 100644 index 00000000..1c802dda --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow/SVM/training_metadata.json @@ -0,0 +1,25 @@ +{ + "preprocessing_steps": [ + "winnow" + ], + "mode": null, + "cv_score": 0.5, + "model_type": "SVC", + "hyperparameters": { + "C": 1, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_eliminate/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_eliminate/RandomForest/model.joblib new file mode 100644 index 00000000..111b80bd Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_eliminate/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_eliminate/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_eliminate/RandomForest/training_metadata.json new file mode 100644 index 00000000..bb670bef --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_eliminate/RandomForest/training_metadata.json @@ -0,0 +1,30 @@ +{ + "preprocessing_steps": [ + "winnow", + "eliminate" + ], + "mode": null, + "cv_score": 0.375, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_eliminate/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_eliminate/SVM/model.joblib new file mode 100644 index 00000000..c0998ade Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_eliminate/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_eliminate/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_eliminate/SVM/training_metadata.json new file mode 100644 index 00000000..525d239f --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_eliminate/SVM/training_metadata.json @@ -0,0 +1,26 @@ +{ + "preprocessing_steps": [ + "winnow", + "eliminate" + ], + "mode": null, + "cv_score": 0.5, + "model_type": "SVC", + "hyperparameters": { + "C": 1, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_eliminate/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_eliminate/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_eliminate/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_eliminate_normalize/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_eliminate_normalize/RandomForest/model.joblib new file mode 100644 index 00000000..bc4e73d7 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_eliminate_normalize/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_eliminate_normalize/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_eliminate_normalize/RandomForest/training_metadata.json new file mode 100644 index 00000000..ffdb5574 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_eliminate_normalize/RandomForest/training_metadata.json @@ -0,0 +1,31 @@ +{ + "preprocessing_steps": [ + "winnow", + "eliminate", + "normalize" + ], + "mode": "both", + "cv_score": 0.3, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_eliminate_normalize/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_eliminate_normalize/SVM/model.joblib new file mode 100644 index 00000000..3e0ab855 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_eliminate_normalize/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_eliminate_normalize/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_eliminate_normalize/SVM/training_metadata.json new file mode 100644 index 00000000..8b221cb6 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_eliminate_normalize/SVM/training_metadata.json @@ -0,0 +1,27 @@ +{ + "preprocessing_steps": [ + "winnow", + "eliminate", + "normalize" + ], + "mode": "both", + "cv_score": 0.425, + "model_type": "SVC", + "hyperparameters": { + "C": 1, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_eliminate_normalize/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_eliminate_normalize/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_eliminate_normalize/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_normalize/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_normalize/RandomForest/model.joblib new file mode 100644 index 00000000..fdf2c8eb Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_normalize/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_normalize/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_normalize/RandomForest/training_metadata.json new file mode 100644 index 00000000..267e47f4 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_normalize/RandomForest/training_metadata.json @@ -0,0 +1,30 @@ +{ + "preprocessing_steps": [ + "winnow", + "normalize" + ], + "mode": "both", + "cv_score": 0.325, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_normalize/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_normalize/SVM/model.joblib new file mode 100644 index 00000000..5a5ae7a4 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_normalize/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_normalize/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_normalize/SVM/training_metadata.json new file mode 100644 index 00000000..a529372d --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_normalize/SVM/training_metadata.json @@ -0,0 +1,26 @@ +{ + "preprocessing_steps": [ + "winnow", + "normalize" + ], + "mode": "both", + "cv_score": 0.4, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_normalize/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_normalize/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_normalize/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_normalize_eliminate/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_normalize_eliminate/RandomForest/model.joblib new file mode 100644 index 00000000..fdf2c8eb Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_normalize_eliminate/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_normalize_eliminate/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_normalize_eliminate/RandomForest/training_metadata.json new file mode 100644 index 00000000..d0cd5505 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_normalize_eliminate/RandomForest/training_metadata.json @@ -0,0 +1,31 @@ +{ + "preprocessing_steps": [ + "winnow", + "normalize", + "eliminate" + ], + "mode": "both", + "cv_score": 0.325, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_normalize_eliminate/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_normalize_eliminate/SVM/model.joblib new file mode 100644 index 00000000..5a5ae7a4 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_normalize_eliminate/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_normalize_eliminate/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_normalize_eliminate/SVM/training_metadata.json new file mode 100644 index 00000000..9abb086a --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_normalize_eliminate/SVM/training_metadata.json @@ -0,0 +1,27 @@ +{ + "preprocessing_steps": [ + "winnow", + "normalize", + "eliminate" + ], + "mode": "both", + "cv_score": 0.4, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_normalize_eliminate/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_normalize_eliminate/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/inner_loop/winnow_normalize_eliminate/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/model.joblib new file mode 100644 index 00000000..3d63aed8 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/training_metadata.json new file mode 100644 index 00000000..77aef16f --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_USF/training_metadata.json @@ -0,0 +1,161 @@ +{ + "site": "USF", + "best_preprocessing_steps": [ + [ + "normalize" + ], + "scale" + ], + "test_score": 0.2, + "model_type": "SVC", + "hyperparameters": { + "C": 1, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + }, + "selected_features": [ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" + ] +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate/RandomForest/model.joblib new file mode 100644 index 00000000..9a515c7c Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate/RandomForest/training_metadata.json new file mode 100644 index 00000000..1e223850 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate/RandomForest/training_metadata.json @@ -0,0 +1,29 @@ +{ + "preprocessing_steps": [ + "eliminate" + ], + "mode": null, + "cv_score": 0.35, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate/SVM/model.joblib new file mode 100644 index 00000000..d02fc0e7 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate/SVM/training_metadata.json new file mode 100644 index 00000000..1f8d2e48 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate/SVM/training_metadata.json @@ -0,0 +1,25 @@ +{ + "preprocessing_steps": [ + "eliminate" + ], + "mode": null, + "cv_score": 0.425, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_normalize/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_normalize/RandomForest/model.joblib new file mode 100644 index 00000000..e9b5c4fb Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_normalize/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_normalize/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_normalize/RandomForest/training_metadata.json new file mode 100644 index 00000000..5a85a401 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_normalize/RandomForest/training_metadata.json @@ -0,0 +1,30 @@ +{ + "preprocessing_steps": [ + "eliminate", + "normalize" + ], + "mode": "both", + "cv_score": 0.3, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_normalize/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_normalize/SVM/model.joblib new file mode 100644 index 00000000..6b3caf01 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_normalize/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_normalize/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_normalize/SVM/training_metadata.json new file mode 100644 index 00000000..43315fa9 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_normalize/SVM/training_metadata.json @@ -0,0 +1,26 @@ +{ + "preprocessing_steps": [ + "eliminate", + "normalize" + ], + "mode": "both", + "cv_score": 0.425, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_normalize/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_normalize/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_normalize/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_normalize_winnow/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_normalize_winnow/RandomForest/model.joblib new file mode 100644 index 00000000..0ae98d8c Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_normalize_winnow/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_normalize_winnow/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_normalize_winnow/RandomForest/training_metadata.json new file mode 100644 index 00000000..432a325b --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_normalize_winnow/RandomForest/training_metadata.json @@ -0,0 +1,31 @@ +{ + "preprocessing_steps": [ + "eliminate", + "normalize", + "winnow" + ], + "mode": "both", + "cv_score": 0.3, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_normalize_winnow/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_normalize_winnow/SVM/model.joblib new file mode 100644 index 00000000..b4579732 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_normalize_winnow/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_normalize_winnow/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_normalize_winnow/SVM/training_metadata.json new file mode 100644 index 00000000..71987408 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_normalize_winnow/SVM/training_metadata.json @@ -0,0 +1,27 @@ +{ + "preprocessing_steps": [ + "eliminate", + "normalize", + "winnow" + ], + "mode": "both", + "cv_score": 0.425, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_normalize_winnow/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_normalize_winnow/selected_features.json new file mode 100644 index 00000000..393bef1b --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_normalize_winnow/selected_features.json @@ -0,0 +1,66 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "hammarbergIndexV_sma3nz_amean", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc2V_sma3nz_amean", + "mfcc3V_sma3nz_amean", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "MeanUnvoicedSegmentLength", + "phonation_ratio", + "mean_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "spectral_gravity", + "spectral_std_dev" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_winnow/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_winnow/RandomForest/model.joblib new file mode 100644 index 00000000..0029cb9f Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_winnow/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_winnow/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_winnow/RandomForest/training_metadata.json new file mode 100644 index 00000000..1adb5c76 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_winnow/RandomForest/training_metadata.json @@ -0,0 +1,30 @@ +{ + "preprocessing_steps": [ + "eliminate", + "winnow" + ], + "mode": null, + "cv_score": 0.425, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_winnow/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_winnow/SVM/model.joblib new file mode 100644 index 00000000..b328c4c3 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_winnow/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_winnow/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_winnow/SVM/training_metadata.json new file mode 100644 index 00000000..0f3b6b2a --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_winnow/SVM/training_metadata.json @@ -0,0 +1,26 @@ +{ + "preprocessing_steps": [ + "eliminate", + "winnow" + ], + "mode": null, + "cv_score": 0.425, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_winnow/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_winnow/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_winnow/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_winnow_normalize/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_winnow_normalize/RandomForest/model.joblib new file mode 100644 index 00000000..b0dd72e1 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_winnow_normalize/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_winnow_normalize/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_winnow_normalize/RandomForest/training_metadata.json new file mode 100644 index 00000000..7e4deca1 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_winnow_normalize/RandomForest/training_metadata.json @@ -0,0 +1,31 @@ +{ + "preprocessing_steps": [ + "eliminate", + "winnow", + "normalize" + ], + "mode": "both", + "cv_score": 0.375, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_winnow_normalize/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_winnow_normalize/SVM/model.joblib new file mode 100644 index 00000000..cd2ff1b0 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_winnow_normalize/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_winnow_normalize/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_winnow_normalize/SVM/training_metadata.json new file mode 100644 index 00000000..70bd6bfc --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_winnow_normalize/SVM/training_metadata.json @@ -0,0 +1,27 @@ +{ + "preprocessing_steps": [ + "eliminate", + "winnow", + "normalize" + ], + "mode": "both", + "cv_score": 0.425, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_winnow_normalize/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_winnow_normalize/selected_features.json new file mode 100644 index 00000000..393bef1b --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/eliminate_winnow_normalize/selected_features.json @@ -0,0 +1,66 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "hammarbergIndexV_sma3nz_amean", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc2V_sma3nz_amean", + "mfcc3V_sma3nz_amean", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "MeanUnvoicedSegmentLength", + "phonation_ratio", + "mean_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "spectral_gravity", + "spectral_std_dev" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize/RandomForest/model.joblib new file mode 100644 index 00000000..4e3d87ee Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize/RandomForest/training_metadata.json new file mode 100644 index 00000000..8a8a7e6e --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize/RandomForest/training_metadata.json @@ -0,0 +1,29 @@ +{ + "preprocessing_steps": [ + "normalize" + ], + "mode": "both", + "cv_score": 0.35, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize/SVM/model.joblib new file mode 100644 index 00000000..536457f3 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize/SVM/training_metadata.json new file mode 100644 index 00000000..64cec636 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize/SVM/training_metadata.json @@ -0,0 +1,25 @@ +{ + "preprocessing_steps": [ + "normalize" + ], + "mode": "both", + "cv_score": 0.425, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_eliminate/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_eliminate/RandomForest/model.joblib new file mode 100644 index 00000000..906713bf Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_eliminate/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_eliminate/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_eliminate/RandomForest/training_metadata.json new file mode 100644 index 00000000..ddda5745 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_eliminate/RandomForest/training_metadata.json @@ -0,0 +1,30 @@ +{ + "preprocessing_steps": [ + "normalize", + "eliminate" + ], + "mode": "both", + "cv_score": 0.2, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_eliminate/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_eliminate/SVM/model.joblib new file mode 100644 index 00000000..cbcd0db7 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_eliminate/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_eliminate/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_eliminate/SVM/training_metadata.json new file mode 100644 index 00000000..93b0cffe --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_eliminate/SVM/training_metadata.json @@ -0,0 +1,26 @@ +{ + "preprocessing_steps": [ + "normalize", + "eliminate" + ], + "mode": "both", + "cv_score": 0.425, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_eliminate/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_eliminate/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_eliminate/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_eliminate_winnow/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_eliminate_winnow/RandomForest/model.joblib new file mode 100644 index 00000000..4a3a6089 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_eliminate_winnow/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_eliminate_winnow/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_eliminate_winnow/RandomForest/training_metadata.json new file mode 100644 index 00000000..2d366176 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_eliminate_winnow/RandomForest/training_metadata.json @@ -0,0 +1,31 @@ +{ + "preprocessing_steps": [ + "normalize", + "eliminate", + "winnow" + ], + "mode": "both", + "cv_score": 0.35, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_eliminate_winnow/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_eliminate_winnow/SVM/model.joblib new file mode 100644 index 00000000..73880abc Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_eliminate_winnow/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_eliminate_winnow/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_eliminate_winnow/SVM/training_metadata.json new file mode 100644 index 00000000..3a3738c9 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_eliminate_winnow/SVM/training_metadata.json @@ -0,0 +1,27 @@ +{ + "preprocessing_steps": [ + "normalize", + "eliminate", + "winnow" + ], + "mode": "both", + "cv_score": 0.425, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_eliminate_winnow/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_eliminate_winnow/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_eliminate_winnow/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_winnow/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_winnow/RandomForest/model.joblib new file mode 100644 index 00000000..c2eb6015 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_winnow/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_winnow/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_winnow/RandomForest/training_metadata.json new file mode 100644 index 00000000..69aea4a3 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_winnow/RandomForest/training_metadata.json @@ -0,0 +1,30 @@ +{ + "preprocessing_steps": [ + "normalize", + "winnow" + ], + "mode": "both", + "cv_score": 0.35, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_winnow/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_winnow/SVM/model.joblib new file mode 100644 index 00000000..568eb39e Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_winnow/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_winnow/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_winnow/SVM/training_metadata.json new file mode 100644 index 00000000..f2e61042 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_winnow/SVM/training_metadata.json @@ -0,0 +1,26 @@ +{ + "preprocessing_steps": [ + "normalize", + "winnow" + ], + "mode": "both", + "cv_score": 0.425, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_winnow/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_winnow/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_winnow/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_winnow_eliminate/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_winnow_eliminate/RandomForest/model.joblib new file mode 100644 index 00000000..c2eb6015 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_winnow_eliminate/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_winnow_eliminate/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_winnow_eliminate/RandomForest/training_metadata.json new file mode 100644 index 00000000..302c2697 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_winnow_eliminate/RandomForest/training_metadata.json @@ -0,0 +1,31 @@ +{ + "preprocessing_steps": [ + "normalize", + "winnow", + "eliminate" + ], + "mode": "both", + "cv_score": 0.35, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_winnow_eliminate/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_winnow_eliminate/SVM/model.joblib new file mode 100644 index 00000000..568eb39e Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_winnow_eliminate/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_winnow_eliminate/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_winnow_eliminate/SVM/training_metadata.json new file mode 100644 index 00000000..b2da4693 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_winnow_eliminate/SVM/training_metadata.json @@ -0,0 +1,27 @@ +{ + "preprocessing_steps": [ + "normalize", + "winnow", + "eliminate" + ], + "mode": "both", + "cv_score": 0.425, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_winnow_eliminate/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_winnow_eliminate/selected_features.json new file mode 100644 index 00000000..393bef1b --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/normalize_winnow_eliminate/selected_features.json @@ -0,0 +1,66 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "hammarbergIndexV_sma3nz_amean", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc2V_sma3nz_amean", + "mfcc3V_sma3nz_amean", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "MeanUnvoicedSegmentLength", + "phonation_ratio", + "mean_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "spectral_gravity", + "spectral_std_dev" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow/RandomForest/model.joblib new file mode 100644 index 00000000..c1755739 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow/RandomForest/training_metadata.json new file mode 100644 index 00000000..980e75b8 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow/RandomForest/training_metadata.json @@ -0,0 +1,29 @@ +{ + "preprocessing_steps": [ + "winnow" + ], + "mode": null, + "cv_score": 0.3, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow/SVM/model.joblib new file mode 100644 index 00000000..abbb1ddf Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow/SVM/training_metadata.json new file mode 100644 index 00000000..d53fabbf --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow/SVM/training_metadata.json @@ -0,0 +1,25 @@ +{ + "preprocessing_steps": [ + "winnow" + ], + "mode": null, + "cv_score": 0.425, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_eliminate/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_eliminate/RandomForest/model.joblib new file mode 100644 index 00000000..c1755739 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_eliminate/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_eliminate/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_eliminate/RandomForest/training_metadata.json new file mode 100644 index 00000000..548d6f08 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_eliminate/RandomForest/training_metadata.json @@ -0,0 +1,30 @@ +{ + "preprocessing_steps": [ + "winnow", + "eliminate" + ], + "mode": null, + "cv_score": 0.3, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_eliminate/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_eliminate/SVM/model.joblib new file mode 100644 index 00000000..abbb1ddf Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_eliminate/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_eliminate/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_eliminate/SVM/training_metadata.json new file mode 100644 index 00000000..258f81ce --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_eliminate/SVM/training_metadata.json @@ -0,0 +1,26 @@ +{ + "preprocessing_steps": [ + "winnow", + "eliminate" + ], + "mode": null, + "cv_score": 0.425, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_eliminate/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_eliminate/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_eliminate/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_eliminate_normalize/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_eliminate_normalize/RandomForest/model.joblib new file mode 100644 index 00000000..0dcdc8c4 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_eliminate_normalize/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_eliminate_normalize/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_eliminate_normalize/RandomForest/training_metadata.json new file mode 100644 index 00000000..3a3833ed --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_eliminate_normalize/RandomForest/training_metadata.json @@ -0,0 +1,31 @@ +{ + "preprocessing_steps": [ + "winnow", + "eliminate", + "normalize" + ], + "mode": "both", + "cv_score": 0.25, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_eliminate_normalize/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_eliminate_normalize/SVM/model.joblib new file mode 100644 index 00000000..8f5ac916 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_eliminate_normalize/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_eliminate_normalize/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_eliminate_normalize/SVM/training_metadata.json new file mode 100644 index 00000000..91b921d4 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_eliminate_normalize/SVM/training_metadata.json @@ -0,0 +1,27 @@ +{ + "preprocessing_steps": [ + "winnow", + "eliminate", + "normalize" + ], + "mode": "both", + "cv_score": 0.425, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_eliminate_normalize/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_eliminate_normalize/selected_features.json new file mode 100644 index 00000000..393bef1b --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_eliminate_normalize/selected_features.json @@ -0,0 +1,66 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "hammarbergIndexV_sma3nz_amean", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc2V_sma3nz_amean", + "mfcc3V_sma3nz_amean", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "MeanUnvoicedSegmentLength", + "phonation_ratio", + "mean_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "spectral_gravity", + "spectral_std_dev" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_normalize/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_normalize/RandomForest/model.joblib new file mode 100644 index 00000000..e6843fe1 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_normalize/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_normalize/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_normalize/RandomForest/training_metadata.json new file mode 100644 index 00000000..0a4776c2 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_normalize/RandomForest/training_metadata.json @@ -0,0 +1,30 @@ +{ + "preprocessing_steps": [ + "winnow", + "normalize" + ], + "mode": "both", + "cv_score": 0.25, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_normalize/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_normalize/SVM/model.joblib new file mode 100644 index 00000000..ee32716b Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_normalize/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_normalize/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_normalize/SVM/training_metadata.json new file mode 100644 index 00000000..0589dcca --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_normalize/SVM/training_metadata.json @@ -0,0 +1,26 @@ +{ + "preprocessing_steps": [ + "winnow", + "normalize" + ], + "mode": "both", + "cv_score": 0.425, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_normalize/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_normalize/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_normalize/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_normalize_eliminate/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_normalize_eliminate/RandomForest/model.joblib new file mode 100644 index 00000000..e6843fe1 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_normalize_eliminate/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_normalize_eliminate/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_normalize_eliminate/RandomForest/training_metadata.json new file mode 100644 index 00000000..e354eddf --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_normalize_eliminate/RandomForest/training_metadata.json @@ -0,0 +1,31 @@ +{ + "preprocessing_steps": [ + "winnow", + "normalize", + "eliminate" + ], + "mode": "both", + "cv_score": 0.25, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_normalize_eliminate/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_normalize_eliminate/SVM/model.joblib new file mode 100644 index 00000000..ee32716b Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_normalize_eliminate/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_normalize_eliminate/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_normalize_eliminate/SVM/training_metadata.json new file mode 100644 index 00000000..f2e1f036 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_normalize_eliminate/SVM/training_metadata.json @@ -0,0 +1,27 @@ +{ + "preprocessing_steps": [ + "winnow", + "normalize", + "eliminate" + ], + "mode": "both", + "cv_score": 0.425, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_normalize_eliminate/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_normalize_eliminate/selected_features.json new file mode 100644 index 00000000..393bef1b --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/inner_loop/winnow_normalize_eliminate/selected_features.json @@ -0,0 +1,66 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "hammarbergIndexV_sma3nz_amean", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc2V_sma3nz_amean", + "mfcc3V_sma3nz_amean", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "MeanUnvoicedSegmentLength", + "phonation_ratio", + "mean_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "spectral_gravity", + "spectral_std_dev" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/model.joblib new file mode 100644 index 00000000..2c5b1e48 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/training_metadata.json new file mode 100644 index 00000000..d466aa6b --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_VUMC/training_metadata.json @@ -0,0 +1,100 @@ +{ + "site": "VUMC", + "best_preprocessing_steps": [ + [ + "normalize", + "winnow", + "eliminate" + ], + "scale" + ], + "test_score": 0.2, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + }, + "selected_features": [ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "hammarbergIndexV_sma3nz_amean", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc2V_sma3nz_amean", + "mfcc3V_sma3nz_amean", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "MeanUnvoicedSegmentLength", + "phonation_ratio", + "mean_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "spectral_gravity", + "spectral_std_dev" + ] +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate/RandomForest/model.joblib new file mode 100644 index 00000000..b429ea3a Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate/RandomForest/training_metadata.json new file mode 100644 index 00000000..3aee8cdd --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate/RandomForest/training_metadata.json @@ -0,0 +1,29 @@ +{ + "preprocessing_steps": [ + "eliminate" + ], + "mode": null, + "cv_score": 0.4, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate/SVM/model.joblib new file mode 100644 index 00000000..feb87afb Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate/SVM/training_metadata.json new file mode 100644 index 00000000..4e7c1607 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate/SVM/training_metadata.json @@ -0,0 +1,25 @@ +{ + "preprocessing_steps": [ + "eliminate" + ], + "mode": null, + "cv_score": 0.375, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_normalize/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_normalize/RandomForest/model.joblib new file mode 100644 index 00000000..949ee622 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_normalize/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_normalize/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_normalize/RandomForest/training_metadata.json new file mode 100644 index 00000000..d0145395 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_normalize/RandomForest/training_metadata.json @@ -0,0 +1,30 @@ +{ + "preprocessing_steps": [ + "eliminate", + "normalize" + ], + "mode": "both", + "cv_score": 0.35, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_normalize/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_normalize/SVM/model.joblib new file mode 100644 index 00000000..30211c3e Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_normalize/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_normalize/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_normalize/SVM/training_metadata.json new file mode 100644 index 00000000..23fd2ca4 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_normalize/SVM/training_metadata.json @@ -0,0 +1,26 @@ +{ + "preprocessing_steps": [ + "eliminate", + "normalize" + ], + "mode": "both", + "cv_score": 0.4, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_normalize/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_normalize/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_normalize/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_normalize_winnow/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_normalize_winnow/RandomForest/model.joblib new file mode 100644 index 00000000..fface47e Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_normalize_winnow/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_normalize_winnow/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_normalize_winnow/RandomForest/training_metadata.json new file mode 100644 index 00000000..88a01f4c --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_normalize_winnow/RandomForest/training_metadata.json @@ -0,0 +1,31 @@ +{ + "preprocessing_steps": [ + "eliminate", + "normalize", + "winnow" + ], + "mode": "both", + "cv_score": 0.35, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_normalize_winnow/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_normalize_winnow/SVM/model.joblib new file mode 100644 index 00000000..d4c8ad51 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_normalize_winnow/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_normalize_winnow/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_normalize_winnow/SVM/training_metadata.json new file mode 100644 index 00000000..80ab93bf --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_normalize_winnow/SVM/training_metadata.json @@ -0,0 +1,27 @@ +{ + "preprocessing_steps": [ + "eliminate", + "normalize", + "winnow" + ], + "mode": "both", + "cv_score": 0.4, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_normalize_winnow/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_normalize_winnow/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_normalize_winnow/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_winnow/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_winnow/RandomForest/model.joblib new file mode 100644 index 00000000..8c4ad2c8 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_winnow/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_winnow/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_winnow/RandomForest/training_metadata.json new file mode 100644 index 00000000..c916f8f3 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_winnow/RandomForest/training_metadata.json @@ -0,0 +1,30 @@ +{ + "preprocessing_steps": [ + "eliminate", + "winnow" + ], + "mode": null, + "cv_score": 0.35, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_winnow/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_winnow/SVM/model.joblib new file mode 100644 index 00000000..976281d1 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_winnow/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_winnow/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_winnow/SVM/training_metadata.json new file mode 100644 index 00000000..add9bc76 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_winnow/SVM/training_metadata.json @@ -0,0 +1,26 @@ +{ + "preprocessing_steps": [ + "eliminate", + "winnow" + ], + "mode": null, + "cv_score": 0.375, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_winnow/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_winnow/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_winnow/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_winnow_normalize/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_winnow_normalize/RandomForest/model.joblib new file mode 100644 index 00000000..57e19da1 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_winnow_normalize/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_winnow_normalize/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_winnow_normalize/RandomForest/training_metadata.json new file mode 100644 index 00000000..43d1e764 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_winnow_normalize/RandomForest/training_metadata.json @@ -0,0 +1,31 @@ +{ + "preprocessing_steps": [ + "eliminate", + "winnow", + "normalize" + ], + "mode": "both", + "cv_score": 0.325, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_winnow_normalize/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_winnow_normalize/SVM/model.joblib new file mode 100644 index 00000000..4e1ca618 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_winnow_normalize/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_winnow_normalize/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_winnow_normalize/SVM/training_metadata.json new file mode 100644 index 00000000..64dd94f0 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_winnow_normalize/SVM/training_metadata.json @@ -0,0 +1,27 @@ +{ + "preprocessing_steps": [ + "eliminate", + "winnow", + "normalize" + ], + "mode": "both", + "cv_score": 0.375, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_winnow_normalize/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_winnow_normalize/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/eliminate_winnow_normalize/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize/RandomForest/model.joblib new file mode 100644 index 00000000..480c0b28 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize/RandomForest/training_metadata.json new file mode 100644 index 00000000..7e3a39c1 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize/RandomForest/training_metadata.json @@ -0,0 +1,29 @@ +{ + "preprocessing_steps": [ + "normalize" + ], + "mode": "both", + "cv_score": 0.325, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize/SVM/model.joblib new file mode 100644 index 00000000..2df28200 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize/SVM/training_metadata.json new file mode 100644 index 00000000..5ec4dfe6 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize/SVM/training_metadata.json @@ -0,0 +1,25 @@ +{ + "preprocessing_steps": [ + "normalize" + ], + "mode": "both", + "cv_score": 0.4, + "model_type": "SVC", + "hyperparameters": { + "C": 1, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_eliminate/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_eliminate/RandomForest/model.joblib new file mode 100644 index 00000000..baebbb8a Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_eliminate/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_eliminate/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_eliminate/RandomForest/training_metadata.json new file mode 100644 index 00000000..c4185a0b --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_eliminate/RandomForest/training_metadata.json @@ -0,0 +1,30 @@ +{ + "preprocessing_steps": [ + "normalize", + "eliminate" + ], + "mode": "both", + "cv_score": 0.25, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_eliminate/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_eliminate/SVM/model.joblib new file mode 100644 index 00000000..6ece7db6 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_eliminate/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_eliminate/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_eliminate/SVM/training_metadata.json new file mode 100644 index 00000000..3fd49d23 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_eliminate/SVM/training_metadata.json @@ -0,0 +1,26 @@ +{ + "preprocessing_steps": [ + "normalize", + "eliminate" + ], + "mode": "both", + "cv_score": 0.375, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_eliminate/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_eliminate/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_eliminate/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_eliminate_winnow/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_eliminate_winnow/RandomForest/model.joblib new file mode 100644 index 00000000..4dd83386 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_eliminate_winnow/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_eliminate_winnow/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_eliminate_winnow/RandomForest/training_metadata.json new file mode 100644 index 00000000..ad05d319 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_eliminate_winnow/RandomForest/training_metadata.json @@ -0,0 +1,31 @@ +{ + "preprocessing_steps": [ + "normalize", + "eliminate", + "winnow" + ], + "mode": "both", + "cv_score": 0.3, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_eliminate_winnow/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_eliminate_winnow/SVM/model.joblib new file mode 100644 index 00000000..7e018214 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_eliminate_winnow/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_eliminate_winnow/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_eliminate_winnow/SVM/training_metadata.json new file mode 100644 index 00000000..f8f6be17 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_eliminate_winnow/SVM/training_metadata.json @@ -0,0 +1,27 @@ +{ + "preprocessing_steps": [ + "normalize", + "eliminate", + "winnow" + ], + "mode": "both", + "cv_score": 0.375, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_eliminate_winnow/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_eliminate_winnow/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_eliminate_winnow/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_winnow/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_winnow/RandomForest/model.joblib new file mode 100644 index 00000000..59da6756 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_winnow/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_winnow/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_winnow/RandomForest/training_metadata.json new file mode 100644 index 00000000..53a27324 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_winnow/RandomForest/training_metadata.json @@ -0,0 +1,30 @@ +{ + "preprocessing_steps": [ + "normalize", + "winnow" + ], + "mode": "both", + "cv_score": 0.3, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_winnow/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_winnow/SVM/model.joblib new file mode 100644 index 00000000..b759d08a Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_winnow/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_winnow/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_winnow/SVM/training_metadata.json new file mode 100644 index 00000000..fa5c42a5 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_winnow/SVM/training_metadata.json @@ -0,0 +1,26 @@ +{ + "preprocessing_steps": [ + "normalize", + "winnow" + ], + "mode": "both", + "cv_score": 0.375, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_winnow/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_winnow/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_winnow/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_winnow_eliminate/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_winnow_eliminate/RandomForest/model.joblib new file mode 100644 index 00000000..59da6756 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_winnow_eliminate/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_winnow_eliminate/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_winnow_eliminate/RandomForest/training_metadata.json new file mode 100644 index 00000000..1d1fe5be --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_winnow_eliminate/RandomForest/training_metadata.json @@ -0,0 +1,31 @@ +{ + "preprocessing_steps": [ + "normalize", + "winnow", + "eliminate" + ], + "mode": "both", + "cv_score": 0.3, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_winnow_eliminate/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_winnow_eliminate/SVM/model.joblib new file mode 100644 index 00000000..b759d08a Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_winnow_eliminate/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_winnow_eliminate/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_winnow_eliminate/SVM/training_metadata.json new file mode 100644 index 00000000..5998c2f4 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_winnow_eliminate/SVM/training_metadata.json @@ -0,0 +1,27 @@ +{ + "preprocessing_steps": [ + "normalize", + "winnow", + "eliminate" + ], + "mode": "both", + "cv_score": 0.375, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_winnow_eliminate/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_winnow_eliminate/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/normalize_winnow_eliminate/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow/RandomForest/model.joblib new file mode 100644 index 00000000..0305f0ac Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow/RandomForest/training_metadata.json new file mode 100644 index 00000000..b0acfd48 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow/RandomForest/training_metadata.json @@ -0,0 +1,29 @@ +{ + "preprocessing_steps": [ + "winnow" + ], + "mode": null, + "cv_score": 0.35, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow/SVM/model.joblib new file mode 100644 index 00000000..ae139635 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow/SVM/training_metadata.json new file mode 100644 index 00000000..73a636b0 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow/SVM/training_metadata.json @@ -0,0 +1,25 @@ +{ + "preprocessing_steps": [ + "winnow" + ], + "mode": null, + "cv_score": 0.4, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_eliminate/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_eliminate/RandomForest/model.joblib new file mode 100644 index 00000000..0305f0ac Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_eliminate/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_eliminate/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_eliminate/RandomForest/training_metadata.json new file mode 100644 index 00000000..eb159e49 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_eliminate/RandomForest/training_metadata.json @@ -0,0 +1,30 @@ +{ + "preprocessing_steps": [ + "winnow", + "eliminate" + ], + "mode": null, + "cv_score": 0.35, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_eliminate/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_eliminate/SVM/model.joblib new file mode 100644 index 00000000..ae139635 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_eliminate/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_eliminate/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_eliminate/SVM/training_metadata.json new file mode 100644 index 00000000..b6bcedb4 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_eliminate/SVM/training_metadata.json @@ -0,0 +1,26 @@ +{ + "preprocessing_steps": [ + "winnow", + "eliminate" + ], + "mode": null, + "cv_score": 0.4, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_eliminate/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_eliminate/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_eliminate/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_eliminate_normalize/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_eliminate_normalize/RandomForest/model.joblib new file mode 100644 index 00000000..8eb0e0a3 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_eliminate_normalize/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_eliminate_normalize/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_eliminate_normalize/RandomForest/training_metadata.json new file mode 100644 index 00000000..2cb2d432 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_eliminate_normalize/RandomForest/training_metadata.json @@ -0,0 +1,31 @@ +{ + "preprocessing_steps": [ + "winnow", + "eliminate", + "normalize" + ], + "mode": "both", + "cv_score": 0.325, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_eliminate_normalize/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_eliminate_normalize/SVM/model.joblib new file mode 100644 index 00000000..3c376658 Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_eliminate_normalize/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_eliminate_normalize/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_eliminate_normalize/SVM/training_metadata.json new file mode 100644 index 00000000..8958c946 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_eliminate_normalize/SVM/training_metadata.json @@ -0,0 +1,27 @@ +{ + "preprocessing_steps": [ + "winnow", + "eliminate", + "normalize" + ], + "mode": "both", + "cv_score": 0.375, + "model_type": "SVC", + "hyperparameters": { + "C": 0.001, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_eliminate_normalize/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_eliminate_normalize/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_eliminate_normalize/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_normalize/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_normalize/RandomForest/model.joblib new file mode 100644 index 00000000..fae61ead Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_normalize/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_normalize/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_normalize/RandomForest/training_metadata.json new file mode 100644 index 00000000..267e47f4 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_normalize/RandomForest/training_metadata.json @@ -0,0 +1,30 @@ +{ + "preprocessing_steps": [ + "winnow", + "normalize" + ], + "mode": "both", + "cv_score": 0.325, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_normalize/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_normalize/SVM/model.joblib new file mode 100644 index 00000000..2c86c7db Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_normalize/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_normalize/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_normalize/SVM/training_metadata.json new file mode 100644 index 00000000..02e8c816 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_normalize/SVM/training_metadata.json @@ -0,0 +1,26 @@ +{ + "preprocessing_steps": [ + "winnow", + "normalize" + ], + "mode": "both", + "cv_score": 0.45, + "model_type": "SVC", + "hyperparameters": { + "C": 1, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_normalize/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_normalize/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_normalize/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_normalize_eliminate/RandomForest/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_normalize_eliminate/RandomForest/model.joblib new file mode 100644 index 00000000..fae61ead Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_normalize_eliminate/RandomForest/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_normalize_eliminate/RandomForest/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_normalize_eliminate/RandomForest/training_metadata.json new file mode 100644 index 00000000..d0cd5505 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_normalize_eliminate/RandomForest/training_metadata.json @@ -0,0 +1,31 @@ +{ + "preprocessing_steps": [ + "winnow", + "normalize", + "eliminate" + ], + "mode": "both", + "cv_score": 0.325, + "model_type": "RandomForestClassifier", + "hyperparameters": { + "bootstrap": true, + "ccp_alpha": 0.0, + "class_weight": null, + "criterion": "gini", + "max_depth": 5, + "max_features": "sqrt", + "max_leaf_nodes": null, + "max_samples": null, + "min_impurity_decrease": 0.0, + "min_samples_leaf": 1, + "min_samples_split": 2, + "min_weight_fraction_leaf": 0.0, + "monotonic_cst": null, + "n_estimators": 10, + "n_jobs": null, + "oob_score": false, + "random_state": 42, + "verbose": 0, + "warm_start": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_normalize_eliminate/SVM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_normalize_eliminate/SVM/model.joblib new file mode 100644 index 00000000..2c86c7db Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_normalize_eliminate/SVM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_normalize_eliminate/SVM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_normalize_eliminate/SVM/training_metadata.json new file mode 100644 index 00000000..57c9f19a --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_normalize_eliminate/SVM/training_metadata.json @@ -0,0 +1,27 @@ +{ + "preprocessing_steps": [ + "winnow", + "normalize", + "eliminate" + ], + "mode": "both", + "cv_score": 0.45, + "model_type": "SVC", + "hyperparameters": { + "C": 1, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + } +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_normalize_eliminate/selected_features.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_normalize_eliminate/selected_features.json new file mode 100644 index 00000000..7d4022e7 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/inner_loop/winnow_normalize_eliminate/selected_features.json @@ -0,0 +1,133 @@ +[ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" +] diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/model.joblib b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/model.joblib new file mode 100644 index 00000000..680ee17b Binary files /dev/null and b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/model.joblib differ diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/training_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/training_metadata.json new file mode 100644 index 00000000..459ebd53 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/site_WCM/training_metadata.json @@ -0,0 +1,161 @@ +{ + "site": "WCM", + "best_preprocessing_steps": [ + [ + "normalize" + ], + "scale" + ], + "test_score": 0.3, + "model_type": "SVC", + "hyperparameters": { + "C": 1, + "break_ties": false, + "cache_size": 200, + "class_weight": null, + "coef0": 0.0, + "decision_function_shape": "ovr", + "degree": 3, + "gamma": "scale", + "kernel": "linear", + "max_iter": -1, + "probability": false, + "random_state": null, + "shrinking": true, + "tol": 0.001, + "verbose": false + }, + "selected_features": [ + "F0semitoneFrom27.5Hz_sma3nz_amean", + "F0semitoneFrom27.5Hz_sma3nz_stddevNorm", + "F0semitoneFrom27.5Hz_sma3nz_percentile20.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile50.0", + "F0semitoneFrom27.5Hz_sma3nz_percentile80.0", + "F0semitoneFrom27.5Hz_sma3nz_pctlrange0-2", + "F0semitoneFrom27.5Hz_sma3nz_meanRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope", + "F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope", + "F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope", + "loudness_sma3_amean", + "loudness_sma3_stddevNorm", + "loudness_sma3_percentile20.0", + "loudness_sma3_percentile50.0", + "loudness_sma3_percentile80.0", + "loudness_sma3_pctlrange0-2", + "loudness_sma3_meanRisingSlope", + "loudness_sma3_stddevRisingSlope", + "loudness_sma3_meanFallingSlope", + "loudness_sma3_stddevFallingSlope", + "spectralFlux_sma3_amean", + "spectralFlux_sma3_stddevNorm", + "mfcc1_sma3_amean", + "mfcc1_sma3_stddevNorm", + "mfcc2_sma3_amean", + "mfcc2_sma3_stddevNorm", + "mfcc3_sma3_amean", + "mfcc3_sma3_stddevNorm", + "mfcc4_sma3_amean", + "mfcc4_sma3_stddevNorm", + "jitterLocal_sma3nz_amean", + "jitterLocal_sma3nz_stddevNorm", + "shimmerLocaldB_sma3nz_amean", + "shimmerLocaldB_sma3nz_stddevNorm", + "HNRdBACF_sma3nz_amean", + "HNRdBACF_sma3nz_stddevNorm", + "logRelF0-H1-H2_sma3nz_amean", + "logRelF0-H1-H2_sma3nz_stddevNorm", + "logRelF0-H1-A3_sma3nz_amean", + "logRelF0-H1-A3_sma3nz_stddevNorm", + "F1frequency_sma3nz_amean", + "F1frequency_sma3nz_stddevNorm", + "F1bandwidth_sma3nz_amean", + "F1bandwidth_sma3nz_stddevNorm", + "F1amplitudeLogRelF0_sma3nz_amean", + "F1amplitudeLogRelF0_sma3nz_stddevNorm", + "F2frequency_sma3nz_amean", + "F2frequency_sma3nz_stddevNorm", + "F2bandwidth_sma3nz_amean", + "F2bandwidth_sma3nz_stddevNorm", + "F2amplitudeLogRelF0_sma3nz_amean", + "F2amplitudeLogRelF0_sma3nz_stddevNorm", + "F3frequency_sma3nz_amean", + "F3frequency_sma3nz_stddevNorm", + "F3bandwidth_sma3nz_amean", + "F3bandwidth_sma3nz_stddevNorm", + "F3amplitudeLogRelF0_sma3nz_amean", + "F3amplitudeLogRelF0_sma3nz_stddevNorm", + "alphaRatioV_sma3nz_amean", + "alphaRatioV_sma3nz_stddevNorm", + "hammarbergIndexV_sma3nz_amean", + "hammarbergIndexV_sma3nz_stddevNorm", + "slopeV0-500_sma3nz_amean", + "slopeV0-500_sma3nz_stddevNorm", + "slopeV500-1500_sma3nz_amean", + "slopeV500-1500_sma3nz_stddevNorm", + "spectralFluxV_sma3nz_amean", + "spectralFluxV_sma3nz_stddevNorm", + "mfcc1V_sma3nz_amean", + "mfcc1V_sma3nz_stddevNorm", + "mfcc2V_sma3nz_amean", + "mfcc2V_sma3nz_stddevNorm", + "mfcc3V_sma3nz_amean", + "mfcc3V_sma3nz_stddevNorm", + "mfcc4V_sma3nz_amean", + "mfcc4V_sma3nz_stddevNorm", + "alphaRatioUV_sma3nz_amean", + "hammarbergIndexUV_sma3nz_amean", + "slopeUV0-500_sma3nz_amean", + "slopeUV500-1500_sma3nz_amean", + "spectralFluxUV_sma3nz_amean", + "loudnessPeaksPerSec", + "VoicedSegmentsPerSec", + "MeanVoicedSegmentLengthSec", + "StddevVoicedSegmentLengthSec", + "MeanUnvoicedSegmentLength", + "StddevUnvoicedSegmentLength", + "equivalentSoundLevel_dBp", + "duration", + "speaking_rate", + "articulation_rate", + "phonation_ratio", + "pause_rate", + "mean_pause_duration", + "mean_f0_hertz", + "std_f0_hertz", + "mean_intensity_db", + "std_intensity_db", + "range_ratio_intensity_db", + "mean_hnr_db", + "std_hnr_db", + "spectral_slope", + "spectral_tilt", + "cepstral_peak_prominence_mean", + "cepstral_peak_prominence_std", + "mean_f1_loc", + "std_f1_loc", + "mean_b1_loc", + "std_b1_loc", + "mean_f2_loc", + "std_f2_loc", + "mean_b2_loc", + "std_b2_loc", + "spectral_gravity", + "spectral_std_dev", + "spectral_skewness", + "spectral_kurtosis", + "local_jitter", + "localabsolute_jitter", + "rap_jitter", + "ppq5_jitter", + "ddp_jitter", + "local_shimmer", + "localDB_shimmer", + "apq3_shimmer", + "apq5_shimmer", + "apq11_shimmer", + "dda_shimmer", + "stoi", + "pesq", + "si_sdr" + ] +} diff --git a/models/audioqc-training-results/training_run_2025-02-21_12-10-09/training_run_metadata.json b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/training_run_metadata.json new file mode 100644 index 00000000..d84006f4 --- /dev/null +++ b/models/audioqc-training-results/training_run_2025-02-21_12-10-09/training_run_metadata.json @@ -0,0 +1,10 @@ +{ + "timestamp": "2025-02-21_12-10-09", + "min_features_to_keep": 64, + "debug_mode": true, + "cv_folds": 5, + "label_column": "label", + "features_csv_path": "/Users/isaacbevers/sensein/b2ai-wrapper/b2ai-data/bridge2ai-voice-corpus-3/derived/static_features.csv", + "participants_tsv_path": "/Users/isaacbevers/sensein/b2ai-wrapper/b2ai-data/bridge2ai-voice-corpus-3/bids/bids/participants.tsv", + "n_jobs": -1 +} diff --git a/src/b2aiprep/audioqc/__init__.py b/src/b2aiprep/audioqc/__init__.py new file mode 100644 index 00000000..e69de29b diff --git a/src/b2aiprep/audioqc/audioqc.py b/src/b2aiprep/audioqc/audioqc.py new file mode 100644 index 00000000..575fa7ba --- /dev/null +++ b/src/b2aiprep/audioqc/audioqc.py @@ -0,0 +1,89 @@ + + +""" +For each site (Outer Loop - Leave-One-Site-Out Cross-Validation): + Hold out one site as the test set + Train on the remaining N-1 sites + + For each hyperparameter configuration (Inner Loop - Grid/Random Search): + Select a combination of: + - Preprocessing method (normalization + feature selection) + - Model type (SVC-lin, SVC-rbf, RFC) + - Classifier hyperparameters (C, gamma, tree depth, etc.) + + For each fold in cross-validation (Inner Loop - Cross-Validation): + Split the training data into train/validation folds + Train the model on the training fold + Evaluate on the validation fold + + Select the best preprocessing/model configuration based on average validation performance + + Train the final model with the best hyperparameters & preprocessing on all N-1 training sites + Evaluate on the held-out site +512 models + +for each site: + hold out one site + training data = remaining sites data + for preprocessing method in [site-wise normalization, filter site origin, winnow] + for center, scale, or both if site-wise normalization: + for model in (SVC-lin, SVC-rbf, RFC) + for model-specific hyperparameters in grid: + run cross-validation on the training data + test the models on the held-out site +pick the best model across all the possible configurations + + + + +## Overall Training Process: +``` +For each site (Outer Loop - Leave-One-Site-Out Cross-Validation): + Hold out one site as the test set + Train on the remaining N-1 sites + + For each hyperparameter configuration (Inner Loop - Grid/Random Search): + Select a combination of: + - Preprocessing method (normalization + feature selection) + - Model type (SVC-lin, SVC-rbf, RFC) + - Classifier hyperparameters (C, gamma, tree depth, etc.) + + For each fold in cross-validation (Inner Loop - Cross-Validation): + Split the training data into train/validation folds + Train the model on the training fold + Evaluate on the validation fold + + Select the best preprocessing/model configuration based on average validation performance + + Train the final model with the best hyperparameters & preprocessing on all N-1 training sites + Evaluate on the held-out site + +``` + + +best_inner_loop_models = [] +for sites: + hold a site out + best_fold_model = inner() + best_inner_loop_models.append(best_fold_model) +best_best_inner_model = best(best_inner_loop_models) +final_model = cross_validation_across_sites(best_best_inner_model) +external_dataset_evaluation(final_model) +""" + + + + +sites = [] +for test_site in sites: + train_sites = sites - test_site + for configuration in configurations: + + + +def check_quality(features_df): + """adds a column with label accept, unsure, or exclude""" + # load model + # classify + # return updated features_df + return features_df diff --git a/src/b2aiprep/audioqc/final_model.joblib b/src/b2aiprep/audioqc/final_model.joblib new file mode 100644 index 00000000..03480d53 Binary files /dev/null and b/src/b2aiprep/audioqc/final_model.joblib differ diff --git a/src/b2aiprep/audioqc/save.py b/src/b2aiprep/audioqc/save.py new file mode 100644 index 00000000..280613dc --- /dev/null +++ b/src/b2aiprep/audioqc/save.py @@ -0,0 +1,49 @@ +import json +import os +from datetime import datetime + +import joblib +import pandas as pd + + +def save_model(output_dir, model, metadata, feature_importance=None, removed_features=None): + """Saves a model and associated metadata.""" + os.makedirs(output_dir, exist_ok=True) + + # Save model + joblib.dump(model, os.path.join(output_dir, "model.joblib")) + + # Save metadata with model type and hyperparameters + metadata.update({"model_type": type(model).__name__, "hyperparameters": model.get_params()}) + + with open(os.path.join(output_dir, "training_metadata.json"), "w") as f: + json.dump(metadata, f, indent=4, ensure_ascii=False) + + # Save feature importance (if available) + if feature_importance is not None: + feature_importance.to_csv(os.path.join(output_dir, "feature_importance.csv"), index=False) + + # Save removed features (if applicable) + if removed_features is not None: + pd.DataFrame(removed_features, columns=["removed_features"]).to_csv( + os.path.join(output_dir, "removed_features.csv"), index=False + ) + + +def save_training_results(base_output_dir, final_model, best_preprocessing_steps, best_model_score): + """Creates structured output directories for all models and metadata.""" + timestamp = datetime.now().strftime("%Y-%m-%d_%H-%M-%S") + run_dir = os.path.join(base_output_dir, f"training_run_{timestamp}") + os.makedirs(run_dir, exist_ok=True) + + # Save final model + save_model( + os.path.join(run_dir, "final_model"), + final_model, + metadata={ + "best_preprocessing_steps": best_preprocessing_steps, + "best_model_score": best_model_score, + }, + ) + + return run_dir diff --git a/src/b2aiprep/audioqc/train.py b/src/b2aiprep/audioqc/train.py new file mode 100755 index 00000000..4665c5ae --- /dev/null +++ b/src/b2aiprep/audioqc/train.py @@ -0,0 +1,701 @@ +"""Implements functions for training AudioQC using a process similar to MRIQC.""" + +import argparse +import json +import logging +import os +from datetime import datetime +from itertools import combinations, permutations + +import numpy as np +import pandas as pd +from joblib import Parallel, delayed +from sklearn.ensemble import ExtraTreesClassifier, RandomForestClassifier +from sklearn.impute import SimpleImputer +from sklearn.model_selection import GridSearchCV, train_test_split +from sklearn.preprocessing import StandardScaler +from sklearn.svm import SVC + +from b2aiprep.audioqc.save import save_model + +logging.basicConfig(level=logging.WARNING) +logger = logging.getLogger(__name__) + +MIN_FEATURES_TO_KEEP = 64 # in feature selection +DEBUG_MODE = True + + +def get_features_df_with_site(features_csv_path, participants_tsv_path): + features_df = pd.read_csv(features_csv_path) + participants_df = pd.read_csv(participants_tsv_path, sep="\t") + participant_to_site = dict(zip(participants_df["record_id"], participants_df["session_site"])) + features_df["site"] = features_df["participant"].map(participant_to_site) + # features_only_df = features_df.drop(columns=['site', 'participant', 'task']) # Exclude non-feature columns + return features_df + + +def site_wise_normalization(features_df, site_column="site", mode="both"): + """ + Perform site-wise normalization using median and interquartile range (IQR). + + Args: + features_df (pd.DataFrame): The dataset including site information. + features (pd.DataFrame): The feature columns to normalize. + site_column (str): The column representing site labels. + mode (str): 'center', 'scale', or 'both' (default). + + Returns: + pd.DataFrame: Normalized feature DataFrame. + """ + non_feature_cols = ["label", "site", "participant", "task"] + # features_only = features_df.drop(columns=non_feature_cols) + features_only = features_df.select_dtypes(include=[np.number]) + + normalized_features = features_only.copy() + + for site in features_df[site_column].unique(): + site_mask = features_df[site_column] == site + site_data = features_only[site_mask] + + median = site_data.median() + iqr = site_data.quantile(0.75) - site_data.quantile(0.25) # Interquartile Range (IQR) + + if mode == "center": + normalized_features.loc[site_mask] = site_data - median + elif mode == "scale": + normalized_features.loc[site_mask] = site_data / iqr + elif mode == "both": + normalized_features.loc[site_mask] = (site_data - median) / iqr + + normalized_features[non_feature_cols] = features_df[non_feature_cols] + normalized_features.replace([np.inf, -np.inf], np.nan, inplace=True) + return normalized_features + + +def site_predictability_feature_elimination( + features_df, max_features_to_remove=100, min_features_to_keep=MIN_FEATURES_TO_KEEP +): + """ + Implements Site-Predictability-Based Feature Elimination using ExtraTreesClassifier. + + This function iteratively removes the most predictive feature for site classification + until either the classifier's performance is near chance level, a predefined + maximum number of features is removed, or the minimum number of features to keep is reached. + + Args: + features_df (pd.DataFrame): DataFrame containing features along with 'site', 'participant', and 'task' columns. + max_features_to_remove (int, optional): Maximum number of features to remove. Default is 67. + min_features_to_keep (int, optional): Minimum number of features to retain. Default is 64. + + Returns: + tuple: + - list: Removed features in order of elimination. + - float: Final site prediction accuracy. + - pd.DataFrame: DataFrame with remaining features after elimination. + """ + # Drop non-AQM columns (site labels are the target variable) + # X = features_df.drop(columns=["site", "participant", "task", "label"]) + X = features_df.select_dtypes(include=[np.number]) + y = features_df["site"] + + # Split data into train and test sets + X_train, X_test, y_train, y_test = train_test_split( + X, y, test_size=0.2, stratify=y, random_state=42 + ) + + # Train initial ExtraTreesClassifier to predict site labels + site_predictor = ExtraTreesClassifier(n_estimators=100, random_state=42) + + if X.isna().sum().sum() > 0: + logger.error( + f"NaN detected before training! Columns with NaNs: {X.columns[X.isna().any()].tolist()}" + ) + logger.error(f"NaN counts: {X.isna().sum()}") + raise ValueError("NaNs detected in training data.") + + site_predictor.fit(X_train, y_train) + + # Get initial accuracy (baseline) + initial_accuracy = site_predictor.score(X_test, y_test) + logger.info(f"Initial site prediction accuracy: {initial_accuracy:.2f}") + + # Define stopping criteria + num_sites = len(np.unique(y)) # Number of unique sites + chance_level = 1 / num_sites # Random guessing accuracy + + # Start feature elimination loop + features_to_remove = [] + iteration = 0 + new_accuracy = 0 + while iteration < max_features_to_remove and len(X_train.columns) > min_features_to_keep: + # Get feature importances + feature_importances = site_predictor.feature_importances_ + + # Identify the most predictive feature + most_predictive_feature = X_train.columns[np.argmax(feature_importances)] + features_to_remove.append(most_predictive_feature) + + # Remove the most predictive feature from the dataset + X_train = X_train.drop(columns=[most_predictive_feature]) + X_test = X_test.drop(columns=[most_predictive_feature]) + + # Retrain the classifier without the removed feature + site_predictor = ExtraTreesClassifier(n_estimators=100, random_state=42) + site_predictor.fit(X_train, y_train) + + # Get new accuracy + new_accuracy = site_predictor.score(X_test, y_test) + logger.info( + f"Iteration {iteration + 1}: Removed '{most_predictive_feature}', New Accuracy: {new_accuracy:.2f}" + ) + + # Check stopping conditions + if new_accuracy <= chance_level: + logger.info("Stopping: Site prediction accuracy is near chance level.") + break + + iteration += 1 + + logger.info(f"Final removed features: {features_to_remove}") + logger.info(f"Final site prediction accuracy: {new_accuracy:.2f}") + + return features_df.drop(columns=features_to_remove) + + +def winnow_feature_selection( + features_df, snr_threshold=1.0, min_features_to_keep=MIN_FEATURES_TO_KEEP +): + """ + Perform Winnow-based feature selection using ExtraTreesClassifier. + + Args: + features_df (pd.DataFrame): DataFrame containing features along with 'site', 'participant', and 'task' columns. + snr_threshold (float): Minimum SNR threshold for feature retention. + min_features_to_keep (int): Minimum number of features to retain. + + Returns: + pd.DataFrame: DataFrame with retained features and original metadata columns. + """ + # Preserve metadata columns + metadata_columns = ["site", "participant", "task", "label"] + + # Extract feature matrix and target labels + X = features_df.select_dtypes(include=[np.number]) + # X = features_df.drop(columns=metadata_columns) + y = features_df["site"] + + # Generate a synthetic random feature (noise) + np.random.seed(42) + X["random_noise"] = np.random.normal(0, 1, size=len(X)) + + # Train ExtraTreesClassifier to measure feature importance + clf = ExtraTreesClassifier(n_estimators=100, random_state=42) + clf.fit(X, y) + + # Get feature importances + feature_importances = clf.feature_importances_ + feature_names = X.columns + + # Identify the importance of the synthetic feature (random noise) + random_feature_importance = feature_importances[X.columns.get_loc("random_noise")] + + # Compute SNR for each feature (ratio of feature importance to random noise importance) + snr = feature_importances / random_feature_importance + + # Select features where SNR exceeds the threshold + features_to_keep = feature_names[snr > snr_threshold].tolist() + + # Avoid including the synthetic "random_noise" feature + if "random_noise" in features_to_keep: + features_to_keep.remove("random_noise") + + # Ensure at least min_features_to_keep are retained + if len(features_to_keep) < min_features_to_keep: + # Sort features by their SNR in descending order + sorted_features = [ + f for f, s in sorted(zip(feature_names, snr), key=lambda x: x[1], reverse=True) + ] + # Exclude the "random_noise" feature from the sorted list + sorted_features = [f for f in sorted_features if f != "random_noise"] + # Select the top features to meet the minimum requirement + features_to_keep = sorted_features[:min_features_to_keep] + + logger.info( + f"Retained {len(features_to_keep)} features after applying SNR threshold and minimum feature constraint." + ) + + # Return the original DataFrame with only selected features and metadata columns + return features_df[metadata_columns + features_to_keep] + + +def add_random_labels( + df, column_name="label", labels=["accept", "exclude", "unsure"], random_seed=None +): + """ + Adds a column with randomly assigned labels to the DataFrame. + + Parameters: + - df (pd.DataFrame): The DataFrame to modify. + - column_name (str): Name of the new column (default: "label"). + - labels (list): List of labels to sample from (default: ["accept", "exclude", "unsure"]). + - random_seed (int or None): Random seed for reproducibility (default: None). + + Returns: + - pd.DataFrame: Modified DataFrame with the new column. + """ + if random_seed is not None: + np.random.seed(random_seed) # Set random seed for reproducibility + + df[column_name] = np.random.choice(labels, size=len(df)) + return df + + +def impute_missing_values(X): + """ + Handles missing values in the dataset by dropping columns that are entirely NaN + and imputing the remaining numeric columns with the mean. + + Args: + X (pd.DataFrame): The feature matrix, including both numeric and non-numeric columns. + + Returns: + pd.DataFrame: The feature matrix with imputed values. + """ + # Separate numeric and non-numeric columns + numeric_X = X.select_dtypes(include=["number"]) + non_numeric_X = X.select_dtypes(include=["object", "string"]) + + # Drop columns that are entirely NaN in numeric data + numeric_X = numeric_X.dropna(axis=1, how="all") + + if numeric_X.isna().sum().sum() > 0: + logger.info("Warning: NaN values detected in numeric features. Imputing missing values...") + imputer = SimpleImputer(strategy="mean") + # Keep original column names + numeric_X = pd.DataFrame(imputer.fit_transform(numeric_X), columns=numeric_X.columns) + + # Recombine numeric and non-numeric data + X_imputed = pd.concat([numeric_X, non_numeric_X], axis=1) + + # **Check if NaNs still remain** + if numeric_X.isna().sum().sum() > 0: + logger.error("NaNs remain AFTER imputation!") + + return X_imputed + + +def preprocess_data(features_df, preprocessing_steps, label_column="label"): + """ + Preprocesses data by handling missing values, standardizing features, + and applying specified preprocessing steps. + + Args: + features_df (pd.DataFrame): DataFrame containing features and labels. + preprocessing_steps (tuple): A tuple containing: + - A set of preprocessing steps to apply ({"normalize", "eliminate", "winnow"}). + - A string representing the normalization mode ("center", "scale", "both", or None). + label_column (str): Name of the column containing classification labels (default: "label"). + + Returns: + tuple: + - X_scaled (np.array): Scaled feature matrix. + - y (pd.Series): Labels. + - selected_features (list): List of selected feature names. + """ + if label_column not in features_df.columns: + raise ValueError(f"Label column '{label_column}' not found in DataFrame.") + + if DEBUG_MODE: + transformed_data = features_df.groupby("site").head(10).copy().reset_index(drop=True) + else: + transformed_data = features_df.copy().reset_index(drop=True) + + steps, mode = preprocessing_steps # Extract preprocessing steps and mode + for step in steps: + if step == "normalize" and mode: + transformed_data = site_wise_normalization(transformed_data, mode=mode) + elif step == "eliminate": + transformed_data = impute_missing_values(transformed_data) + transformed_data = site_predictability_feature_elimination(transformed_data) + elif step == "winnow": + transformed_data = impute_missing_values(transformed_data) + transformed_data = winnow_feature_selection(transformed_data) + + # Identify columns that are entirely NaN and replace only those with zeroes + transformed_data.replace([np.inf, -np.inf], np.nan, inplace=True) + nan_columns = transformed_data.columns[transformed_data.isna().all()] + transformed_data[nan_columns] = 0 + selected_features = transformed_data.select_dtypes(include=["number"]).columns.tolist() + X = transformed_data[selected_features] # Keep only selected features + y = transformed_data[label_column] + + logger.info(f"Selected {len(selected_features)} features for training.") + X = impute_missing_values(X) + + # Standardize features + scaler = StandardScaler() + X_scaled = scaler.fit_transform(X) + + return X_scaled, y, selected_features + + +def svm_train(X, y, cv_folds=5): + """ + Trains and evaluates Support Vector Machines (SVM) using cross-validation. + + Args: + X (pd.DataFrame or np.array): Feature matrix. + y (pd.Series or np.array): Labels. + cv_folds (int): Number of cross-validation folds. + + Returns: + dict: Contains the best SVM model and its cross-validation score. + """ + if DEBUG_MODE: + # Minimal search space for speed + param_grid_svm = { + "C": [1e-3, 1], + "kernel": ["linear"], # Only one kernel + "gamma": ["scale"], # Single gamma setting + } + else: + # Larger search space + param_grid_svm = { + "C": [1e-3, 1e-1, 1, 10, 100], + "kernel": ["linear", "poly", "rbf"], + "gamma": ["scale", 1e-2, 1e-1, 1], + "degree": [2, 3, 4], + "coef0": [0.0, 0.5], + "class_weight": ["balanced"], + } + + grid_search = GridSearchCV(SVC(), param_grid_svm, cv=cv_folds, scoring="accuracy", n_jobs=-1) + grid_search.fit(X, y) + + best_svc = grid_search.best_estimator_ + best_params = grid_search.best_params_ + best_score = grid_search.best_score_ + + logger.info(f"Best SVM {best_params}, CV Accuracy: {best_score:.4f}") + + return { + "best_svc": best_svc, + "best_params": best_params, + "cv_accuracy": best_score, + } + + +def rfc_train(X, y, cv_folds=5): + """ + Trains and evaluates a Random Forest Classifier (RFC) using cross-validation. + + Args: + X (pd.DataFrame or np.array): Feature matrix. + y (pd.Series or np.array): Labels. + cv_folds (int): Number of cross-validation folds. + + Returns: + dict: Contains the best RFC model and its cross-validation score. + """ + if DEBUG_MODE: + param_grid_rfc = { + "n_estimators": [10], + "max_depth": [5], + "min_samples_split": [2], + "min_samples_leaf": [1], + "max_features": ["sqrt"], + "bootstrap": [True], + "criterion": ["gini"], + } + else: + param_grid_rfc = { + "n_estimators": [100, 500], + "max_depth": [None, 10, 30], + "min_samples_split": [2, 10], + "min_samples_leaf": [1, 2, 4], + "max_features": ["auto", "sqrt", "log2"], + "bootstrap": [True], + "criterion": ["gini", "entropy"], + "class_weight": ["balanced"], + } + + grid_search = GridSearchCV( + RandomForestClassifier(random_state=42), + param_grid_rfc, + cv=cv_folds, + scoring="accuracy", + n_jobs=-1, + ) + grid_search.fit(X, y) + + best_rfc = grid_search.best_estimator_ + best_params = grid_search.best_params_ + best_score = grid_search.best_score_ + + logger.info(f"Best Random Forest {best_params}, CV Accuracy: {best_score:.4f}") + + return { + "best_rfc": best_rfc, + "best_params": best_params, + "cv_accuracy": best_score, + } + + +def train_final_model( + features_df, best_inner_model, best_preprocessing_steps, label_column="label" +): + """ + Applies the best preprocessing steps to the entire dataset in the exact order + and trains a new model on all data using the best hyperparameters. + + Args: + features_df (pd.DataFrame): The complete dataset with features and labels. + best_inner_model (sklearn estimator): The best-performing model from the inner loop. + best_preprocessing_steps (tuple): The preprocessing steps used by the best model. + label_column (str): Name of the classification label column. + + Returns: + final_model: The newly trained model using the best hyperparameters on all data. + """ + X_all, y_all, selected_features = preprocess_data( + features_df, best_preprocessing_steps, label_column=label_column + ) + + best_model_class = type(best_inner_model) + best_model_params = best_inner_model.get_params() + final_model = best_model_class(**best_model_params) + final_model.fit(X_all, y_all) + return final_model, selected_features + + +def inner_loop(features_df, label_column="label", cv_folds=5, output_dir="training_results"): + """Performs an inner-loop search over feature preprocessing configurations, + trains SVM and RFC models, and selects the best-performing model. + + Args: + features_df (pd.DataFrame): DataFrame containing features and labels. + label_column (str): Name of the classification label column. + cv_folds (int): Number of cross-validation folds. + output_dir (str): Directory where model results are saved. + + Returns: + tuple: + - best_model (sklearn estimator): The best-performing model. + - best_score (float): The best cross-validation accuracy score. + - best_steps (tuple): The preprocessing steps that yielded the best-performing model. + - selected_features (list): List of selected feature column names. + """ + best_model = None + best_score = 0 + best_steps = None + selected_features = None # To store feature names for best model + + preprocessing_steps = ["normalize", "eliminate", "winnow"] + preprocessing_permutations = [ + list(permutation) + for r in range(1, len(preprocessing_steps) + 1) + for combination in combinations(preprocessing_steps, r) + for permutation in permutations(combination) + ] + + # if DEBUG_MODE: + # preprocessing_permutations = [["normalize"]] + + for preprocessing_permutation in preprocessing_permutations: + normalize_modes = ( + ["center", "scale", "both"] if "normalize" in preprocessing_permutation else [None] + ) + + for mode in normalize_modes: + transformed_data = features_df.copy().reset_index(drop=True) + + X, y, selected_feature_names = preprocess_data( + transformed_data, (preprocessing_permutation, mode), label_column + ) + + svm_results = svm_train(X, y, cv_folds=cv_folds) + rfc_results = rfc_train(X, y, cv_folds=cv_folds) + + for model_type, results in [ + ("SVM", svm_results), + ("RandomForest", rfc_results), + ]: + model = results["best_svc"] if model_type == "SVM" else results["best_rfc"] + score = results["cv_accuracy"] + + # Save model for this preprocessing step + step_dir = os.path.join( + output_dir, "inner_loop", "_".join(preprocessing_permutation) + ) + os.makedirs(step_dir, exist_ok=True) + + # Save selected features + feature_file = os.path.join(step_dir, "selected_features.json") + with open(feature_file, "w") as f: + json.dump(selected_features, f, indent=4) + + save_model( + os.path.join(step_dir, model_type), + model, + metadata={ + "preprocessing_steps": preprocessing_permutation, + "mode": mode, + "cv_score": score, + "model_type": type(model).__name__, + "hyperparameters": model.get_params(), + }, + ) + + if score > best_score: + best_model = model + best_score = score + best_steps = (preprocessing_permutation, mode) + selected_features = ( + selected_feature_names # Update selected features for best model + ) + + return best_model, best_score, best_steps, selected_features + + +def train_qc_classifier( + features_csv_path, + participants_tsv_path, + label_column="label", + cv_folds=5, + base_output_dir="training_results", + n_jobs=-1, # Enables parallel processing with automatic CPU allocation +): + """Performs an outer-loop cross-validation process using a leave-one-site-out (LoSo) approach. + Trains models using the inner loop, selects the best-performing model across all site folds, + and then trains a new model with the best hyperparameters on all data. + + Args: + features_csv_path (str): Path to the features CSV file. + participants_tsv_path (str): Path to the participants TSV file. + label_column (str): Name of the classification label column. + cv_folds (int): Number of cross-validation folds. + base_output_dir (str): Output directory for saved models. + n_jobs (int): Number of parallel jobs (-1 = all available CPUs). + + Returns: + final_model: The trained model using the best hyperparameters. + """ + # Create a unique directory for this training run + timestamp = datetime.now().strftime("%Y-%m-%d_%H-%M-%S") + run_dir = os.path.join(base_output_dir, f"training_run_{timestamp}") + os.makedirs(run_dir, exist_ok=True) + + # Save training metadata + training_run_metadata = { + "timestamp": timestamp, + "min_features_to_keep": MIN_FEATURES_TO_KEEP, + "debug_mode": DEBUG_MODE, + "cv_folds": cv_folds, + "label_column": label_column, + "features_csv_path": features_csv_path, + "participants_tsv_path": participants_tsv_path, + "n_jobs": n_jobs, + } + with open(os.path.join(run_dir, "training_run_metadata.json"), "w") as f: + json.dump(training_run_metadata, f, indent=4, ensure_ascii=False) + + # Load and process data + features_df = get_features_df_with_site(features_csv_path, participants_tsv_path) + features_df = add_random_labels(features_df) + unique_sites = features_df["site"].unique() + + # Define function for parallel execution + def process_site(site): + """Handles inner loop training for a single site.""" + logger.info(f"Processing site: {site}") + + train_df = features_df[features_df["site"] != site].copy().reset_index(drop=True) + test_df = features_df[features_df["site"] == site].copy().reset_index(drop=True) + site_dir = os.path.join(run_dir, f"site_{site}") + os.makedirs(site_dir, exist_ok=True) + + # Run inner loop training + best_fold_model, best_fold_score, best_fold_steps, selected_features = inner_loop( + train_df, label_column, cv_folds, site_dir + ) + + # Apply best preprocessing to test set + test_steps = ( + (["normalize"], best_fold_steps[1]) if "normalize" in best_fold_steps[0] else ([], None) + ) + test_df = test_df[selected_features + [label_column, "site", "participant", "task"]] + X_test, y_test, _ = preprocess_data(test_df, test_steps, label_column) + best_model_score = best_fold_model.score(X_test, y_test) + + logger.info( + f"Best model for site {site}: {best_fold_model} with test score {best_model_score:.4f}" + ) + + # Save best model for site + save_model( + site_dir, + best_fold_model, + metadata={ + "site": site, + "best_preprocessing_steps": best_fold_steps, + "test_score": best_model_score, + "model_type": type(best_fold_model).__name__, + "hyperparameters": best_fold_model.get_params(), + "selected_features": selected_features, + }, + ) + + return best_fold_model, best_fold_score, best_fold_steps + + # Run inner loop training in parallel across sites + best_inner_loop_models = Parallel(n_jobs=n_jobs)( + delayed(process_site)(site) for site in unique_sites + ) + + # Select the best model from all sites + best_inner_model, best_model_score, best_preprocessing_steps = max( + best_inner_loop_models, key=lambda x: x[1] + ) + logger.info(f"Best model selected: {best_inner_model} with score {best_model_score:.4f}") + + # Train final model with best parameters + final_model, selected_features = train_final_model( + features_df, best_inner_model, best_preprocessing_steps, label_column + ) + + # Save final model + final_model_dir = os.path.join(run_dir, "final_model") + os.makedirs(final_model_dir, exist_ok=True) + save_model( + final_model_dir, + final_model, + metadata={ + "best_preprocessing_steps": best_preprocessing_steps, + "best_model_score": best_model_score, + "model_type": type(final_model).__name__, + "hyperparameters": final_model.get_params(), + "selected_features": selected_features, + }, + ) + + logger.info(f"Training run saved at: {run_dir}") + return final_model + + +if __name__ == "__main__": + parser = argparse.ArgumentParser(description="Train AudioQC model using MRIQC-like pipeline.") + parser.add_argument("--features_csv", required=True, help="Path to the features CSV file.") + parser.add_argument( + "--participants_tsv", required=True, help="Path to the participants TSV file." + ) + parser.add_argument( + "--output_dir", required=True, help="Base directory for saving training results." + ) + + args = parser.parse_args() + + train_qc_classifier( + features_csv_path=args.features_csv, + participants_tsv_path=args.participants_tsv, + base_output_dir=args.output_dir, + ) diff --git a/src/b2aiprep/cli.py b/src/b2aiprep/cli.py index b2519032..9d04bf68 100644 --- a/src/b2aiprep/cli.py +++ b/src/b2aiprep/cli.py @@ -21,7 +21,7 @@ @click.group() @click.option( "--log-level", - default="INFO", + default="WARNING", type=click.Choice(["DEBUG", "INFO", "WARNING", "ERROR", "CRITICAL"]), help="Set the log level for the CLI.", ) @@ -50,6 +50,7 @@ def cli(ctx, log_level): cli.add_command(reproschema_audio_to_folder) cli.add_command(reproschema_to_redcap) + if __name__ == "__main__": # include main to enable python debugging cli() # pylint: disable=no-value-for-parameter diff --git a/src/b2aiprep/commands.py b/src/b2aiprep/commands.py index c966a10e..57d39754 100644 --- a/src/b2aiprep/commands.py +++ b/src/b2aiprep/commands.py @@ -1,25 +1,22 @@ """Commands available through the CLI.""" import csv -from functools import partial import json import logging import os import shutil -import tarfile -import typing as t +from functools import partial from glob import glob -from pathlib import Path from importlib import resources from collections import OrderedDict from b2aiprep.prepare.derived_data import feature_extraction_generator, spectrogram_generator import click -import numpy as np import pandas as pd import pkg_resources import pydra import torch from datasets import Dataset +from pyarrow.parquet import SortingColumn from pydra.mark import annotate from senselab.audio.data_structures.audio import Audio from senselab.audio.tasks.features_extraction.opensmile import ( @@ -50,7 +47,6 @@ from b2aiprep.prepare.prepare import extract_features_workflow, extract_features_sequentially,validate_bids_data, clean_phenotype_data from b2aiprep.prepare.reproschema_to_redcap import parse_survey, parse_audio - # from b2aiprep.synthetic_data import generate_synthetic_tabular_data _LOGGER = logging.getLogger(__name__) @@ -97,6 +93,7 @@ def redcap2bids( ) + @click.command() @click.argument("bids_dir_path", type=click.Path()) @click.option("--redcap_csv_path", type=click.Path(exists=True), default=None, show_default=True) @@ -188,17 +185,18 @@ def prepare_bids( ) _LOGGER.info("Audio feature extraction complete.") - if validate: - # Below code checks to see if we have all the expected feature/transcript files. - validate_bids_folder(Path(bids_dir_path)) +# if validate: +# # Below code checks to see if we have all the expected feature/transcript files. +# validate_bids_folder(Path(bids_dir_path)) - if tar_file_path is not None: - _LOGGER.info("Saving .tar file with processed data...") - with tarfile.open(tar_file_path, "w:gz") as tar: - tar.add(bids_dir_path, arcname=os.path.basename(bids_dir_path)) - _LOGGER.info(f"Saved processed data .tar file at: {tar_file_path}") +# if tar_file_path is not None: +# _LOGGER.info("Saving .tar file with processed data...") +# with tarfile.open(tar_file_path, "w:gz") as tar: +# tar.add(bids_dir_path, arcname=os.path.basename(bids_dir_path)) +# _LOGGER.info(f"Saved processed data .tar file at: {tar_file_path}") + +# _LOGGER.info("Process completed.") - _LOGGER.info("Process completed.") @click.command() @@ -235,7 +233,7 @@ def create_derived_dataset(bids_path, outdir): audio_paths = sorted( audio_paths, # sort first by subject, then by task - key=lambda x: (x.stem.split('_')[0], x.stem.split('_')[2]) + key=lambda x: (x.stem.split("_")[0], x.stem.split("_")[2]), ) # remove known subjects without any audio @@ -249,8 +247,8 @@ def create_derived_dataset(bids_path, outdir): _LOGGER.info("Loading spectrograms into a single HF dataset.") - for feature_name in ['spectrogram', 'mfcc']: - if feature_name == 'mfcc': + for feature_name in ["spectrogram", "mfcc"]: + if feature_name == "mfcc": use_byte_stream_split = True audio_feature_generator = partial( feature_extraction_generator, @@ -271,8 +269,10 @@ def create_derived_dataset(bids_path, outdir): version="2.6", compression="zstd", # Better compression ratio than snappy, still good speed compression_level=3, + # Enable dictionary encoding for strings use_dictionary=["participant_id", "session_id", "task_name"], + write_statistics=True, # enable page index for better filtering data_page_size=1_048_576, # 1MB pages @@ -310,9 +310,11 @@ def create_derived_dataset(bids_path, outdir): transcription = features.get("transcription", None) if transcription is not None: transcription = transcription.text + if subj_info['task_name'].lower().startswith('free-speech') or \ subj_info['task_name'].lower().startswith('audio-check') or \ subj_info['task_name'].lower().startswith('open-response-questions'): + # we omit tasks where free speech occurs transcription = None subj_info["transcription"] = transcription @@ -339,7 +341,7 @@ def create_derived_dataset(bids_path, outdir): df = pd.read_csv(bids_path.joinpath("participants.tsv"), sep="\t") # remove subject - idx = df['record_id'].isin(SUBJECTS_TO_REMOVE) + idx = df["record_id"].isin(SUBJECTS_TO_REMOVE) if idx.sum() > 0: _LOGGER.info( f"Removing {idx.sum()} records from phenotype due to hard-coded subject removal.")