readme.txt

This is the runbook for the project code

Code on GitHub.com/dpnolan/voxpop

Input and output data on Google Drive

Input and output data on Google Drive

Documentation

Report \ pnol22154116.pdf

Video Presentation on Google Drive

Directory structure

\voxpop

Scripts should run in the home directory

\ANT

\DUB

Hold the input WAV files, 50-100 seconds of a single speaker reading out scripted sentences, the same for everyone ANT and DUB subdirectories hold the recordings coming directly from SAIE in .wav format Fed into the preprocessing scripts

sample_output_files

Output from preprocessing scripts i.e. DataFrame pickles of personal data and the sound file, MFCC, MFCC delta and MFCC delta 2 Sample files are the WAV sound files coming from taking one second-long samples of the full speaker's sound recording

Scripts

Requirements files are renamed with the Jupyter kernel they were extracted from

Pre DUB / ANTBEL $0 73$

Jupyter notebook with Python Preprocesing of the input files to produce the sample files and the DataFrame pickle files

clustering12.ipynb

k-means clustering models Logistic regressions Fed by the sample files and the pickle files

EDA.ipynb

Exploratory Data Analysis Fed by the sample files and the pickle files

TF8 MLP.ipynb

TF MLP Neural network Fed by the sample files and the pickle files

TF10TK CNN.ipynb

TF CNN Convolutional neural network Fed by the sample files and the pickle files