Lipnet_model.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "gpuType": "T4"
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU",
    "gpuClass": "standard"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "source": [
        "## Install and Imports"
      ],
      "metadata": {
        "id": "T86ZQxxED6Ws"
      }
    },
    {
      "cell_type": "code",
      "execution_count": 155,
      "metadata": {
        "id": "5raaYBXofWvH",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "e72f476d-49cd-4836-8254-0ea0bae6a9b4"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Looking in indexes: https://pypi.org/simple, https://us-python.pkg.dev/colab-wheels/public/simple/\n",
            "Requirement already satisfied: gdown in /usr/local/lib/python3.10/dist-packages (4.6.6)\n",
            "Requirement already satisfied: imageio in /usr/local/lib/python3.10/dist-packages (2.25.1)\n",
            "Requirement already satisfied: filelock in /usr/local/lib/python3.10/dist-packages (from gdown) (3.12.0)\n",
            "Requirement already satisfied: requests[socks] in /usr/local/lib/python3.10/dist-packages (from gdown) (2.27.1)\n",
            "Requirement already satisfied: six in /usr/local/lib/python3.10/dist-packages (from gdown) (1.16.0)\n",
            "Requirement already satisfied: tqdm in /usr/local/lib/python3.10/dist-packages (from gdown) (4.65.0)\n",
            "Requirement already satisfied: beautifulsoup4 in /usr/local/lib/python3.10/dist-packages (from gdown) (4.11.2)\n",
            "Requirement already satisfied: numpy in /usr/local/lib/python3.10/dist-packages (from imageio) (1.22.4)\n",
            "Requirement already satisfied: pillow>=8.3.2 in /usr/local/lib/python3.10/dist-packages (from imageio) (8.4.0)\n",
            "Requirement already satisfied: soupsieve>1.2 in /usr/local/lib/python3.10/dist-packages (from beautifulsoup4->gdown) (2.4.1)\n",
            "Requirement already satisfied: urllib3<1.27,>=1.21.1 in /usr/local/lib/python3.10/dist-packages (from requests[socks]->gdown) (1.26.15)\n",
            "Requirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.10/dist-packages (from requests[socks]->gdown) (2022.12.7)\n",
            "Requirement already satisfied: charset-normalizer~=2.0.0 in /usr/local/lib/python3.10/dist-packages (from requests[socks]->gdown) (2.0.12)\n",
            "Requirement already satisfied: idna<4,>=2.5 in /usr/local/lib/python3.10/dist-packages (from requests[socks]->gdown) (3.4)\n",
            "Requirement already satisfied: PySocks!=1.5.7,>=1.5.6 in /usr/local/lib/python3.10/dist-packages (from requests[socks]->gdown) (1.7.1)\n"
          ]
        }
      ],
      "source": [
        "!pip install gdown imageio"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import os\n",
        "import cv2\n",
        "import tensorflow as tf\n",
        "import gdown #get data from GDrive\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "import imageio #render post processed output a lot better (better while working w/ videos)\n",
        "from typing import List #List type annotations"
      ],
      "metadata": {
        "id": "pEtg8rIpDv9k"
      },
      "execution_count": 156,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "tf.config.list_physical_devices('GPU')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "y3ogQiAaDz8f",
        "outputId": "fd1e6cac-5be4-4451-b519-3a3c5c079701"
      },
      "execution_count": 157,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[PhysicalDevice(name='/physical_device:GPU:0', device_type='GPU')]"
            ]
          },
          "metadata": {},
          "execution_count": 157
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#prevents machine from sucking up all memory and get out of memory errors\n",
        "physical_devices = tf.config.list_physical_devices('GPU')\n",
        "try:\n",
        "  tf.config.experimental.set_memory_growth(physical_devices[0],True)\n",
        "except:\n",
        "  pass"
      ],
      "metadata": {
        "id": "pA-oII3ID298"
      },
      "execution_count": 158,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Build Data Loading Functions:"
      ],
      "metadata": {
        "id": "IWd3RzQhEEGF"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "The dataset used is an extract of the original Grid Corpus dataset. Had specific videos/ part of the dataset (i.e. only taken videos and transcripts from a single speaker, the original dataset has around 34 speakers) loaded into gdrive, would import the dataset from there.\n",
        "\n",
        "The Grid Corpus is a large multitalker audiovisual sentence corpus designed to support joint computational-behavioral studies in speech perception. In brief, the corpus consists of high-quality audio and video (facial) recordings of 1000 sentences spoken by each of 34 talkers (18 male, 16 female), for a total of 34000 sentences. Sentences are of the form \"put red at G9 now\".\n",
        "\n",
        "*More about the original dataset: [Grid Corpus](https://zenodo.org/record/3625687#.ZFvXhXZBxEY)*"
      ],
      "metadata": {
        "id": "9WHrNXuAEJYd"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "url= \"https://drive.google.com/uc?id=1YlvpDLix3S-U8fd-gqRwPcWXAXm8JwjL\"\n",
        "output = 'data.zip'\n",
        "gdown.download(url,output,quiet=False)\n",
        "gdown.extractall('data.zip')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "AZTdenOYD49Y",
        "outputId": "999fa6d4-b1dd-46b5-cbb0-bfab2f1a4005"
      },
      "execution_count": 159,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "Downloading...\n",
            "From: https://drive.google.com/uc?id=1YlvpDLix3S-U8fd-gqRwPcWXAXm8JwjL\n",
            "To: /content/data.zip\n",
            "100%|██████████| 423M/423M [00:06<00:00, 63.0MB/s]\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['data/',\n",
              " 'data/alignments/',\n",
              " 'data/alignments/s1/',\n",
              " 'data/alignments/s1/bbaf2n.align',\n",
              " 'data/alignments/s1/bbaf3s.align',\n",
              " 'data/alignments/s1/bbaf4p.align',\n",
              " 'data/alignments/s1/bbaf5a.align',\n",
              " 'data/alignments/s1/bbal6n.align',\n",
              " 'data/alignments/s1/bbal7s.align',\n",
              " 'data/alignments/s1/bbal8p.align',\n",
              " 'data/alignments/s1/bbal9a.align',\n",
              " 'data/alignments/s1/bbas1s.align',\n",
              " 'data/alignments/s1/bbas2p.align',\n",
              " 'data/alignments/s1/bbas3a.align',\n",
              " 'data/alignments/s1/bbaszn.align',\n",
              " 'data/alignments/s1/bbaz4n.align',\n",
              " 'data/alignments/s1/bbaz5s.align',\n",
              " 'data/alignments/s1/bbaz6p.align',\n",
              " 'data/alignments/s1/bbaz7a.align',\n",
              " 'data/alignments/s1/bbbf6n.align',\n",
              " 'data/alignments/s1/bbbf7s.align',\n",
              " 'data/alignments/s1/bbbf8p.align',\n",
              " 'data/alignments/s1/bbbf9a.align',\n",
              " 'data/alignments/s1/bbbm1s.align',\n",
              " 'data/alignments/s1/bbbm2p.align',\n",
              " 'data/alignments/s1/bbbm3a.align',\n",
              " 'data/alignments/s1/bbbmzn.align',\n",
              " 'data/alignments/s1/bbbs4n.align',\n",
              " 'data/alignments/s1/bbbs5s.align',\n",
              " 'data/alignments/s1/bbbs6p.align',\n",
              " 'data/alignments/s1/bbbs7a.align',\n",
              " 'data/alignments/s1/bbbz8n.align',\n",
              " 'data/alignments/s1/bbbz9s.align',\n",
              " 'data/alignments/s1/bbie8n.align',\n",
              " 'data/alignments/s1/bbie9s.align',\n",
              " 'data/alignments/s1/bbif1a.align',\n",
              " 'data/alignments/s1/bbifzp.align',\n",
              " 'data/alignments/s1/bbil2n.align',\n",
              " 'data/alignments/s1/bbil3s.align',\n",
              " 'data/alignments/s1/bbil4p.align',\n",
              " 'data/alignments/s1/bbil5a.align',\n",
              " 'data/alignments/s1/bbir6n.align',\n",
              " 'data/alignments/s1/bbir7s.align',\n",
              " 'data/alignments/s1/bbir8p.align',\n",
              " 'data/alignments/s1/bbir9a.align',\n",
              " 'data/alignments/s1/bbiz1s.align',\n",
              " 'data/alignments/s1/bbiz2p.align',\n",
              " 'data/alignments/s1/bbiz3a.align',\n",
              " 'data/alignments/s1/bbizzn.align',\n",
              " 'data/alignments/s1/bbwg1s.align',\n",
              " 'data/alignments/s1/bbwg2p.align',\n",
              " 'data/alignments/s1/bbwg3a.align',\n",
              " 'data/alignments/s1/bbwgzn.align',\n",
              " 'data/alignments/s1/bbwm4n.align',\n",
              " 'data/alignments/s1/bbwm5s.align',\n",
              " 'data/alignments/s1/bbwm6p.align',\n",
              " 'data/alignments/s1/bbwm7a.align',\n",
              " 'data/alignments/s1/bbws8n.align',\n",
              " 'data/alignments/s1/bbws9s.align',\n",
              " 'data/alignments/s1/bbwt1a.align',\n",
              " 'data/alignments/s1/bbwtzp.align',\n",
              " 'data/alignments/s1/bgaa6n.align',\n",
              " 'data/alignments/s1/bgaa7s.align',\n",
              " 'data/alignments/s1/bgaa8p.align',\n",
              " 'data/alignments/s1/bgaa9a.align',\n",
              " 'data/alignments/s1/bgah1s.align',\n",
              " 'data/alignments/s1/bgah2p.align',\n",
              " 'data/alignments/s1/bgah3a.align',\n",
              " 'data/alignments/s1/bgahzn.align',\n",
              " 'data/alignments/s1/bgan4n.align',\n",
              " 'data/alignments/s1/bgan5s.align',\n",
              " 'data/alignments/s1/bgan6p.align',\n",
              " 'data/alignments/s1/bgan7a.align',\n",
              " 'data/alignments/s1/bgat8n.align',\n",
              " 'data/alignments/s1/bgat9s.align',\n",
              " 'data/alignments/s1/bgau1a.align',\n",
              " 'data/alignments/s1/bgauzp.align',\n",
              " 'data/alignments/s1/bgbb1s.align',\n",
              " 'data/alignments/s1/bgbb2p.align',\n",
              " 'data/alignments/s1/bgbb3a.align',\n",
              " 'data/alignments/s1/bgbbzn.align',\n",
              " 'data/alignments/s1/bgbh4n.align',\n",
              " 'data/alignments/s1/bgbh5s.align',\n",
              " 'data/alignments/s1/bgbh6p.align',\n",
              " 'data/alignments/s1/bgbh7a.align',\n",
              " 'data/alignments/s1/bgbn8n.align',\n",
              " 'data/alignments/s1/bgbn9s.align',\n",
              " 'data/alignments/s1/bgbo1a.align',\n",
              " 'data/alignments/s1/bgbozp.align',\n",
              " 'data/alignments/s1/bgbu2n.align',\n",
              " 'data/alignments/s1/bgbu3s.align',\n",
              " 'data/alignments/s1/bgbu4p.align',\n",
              " 'data/alignments/s1/bgbu5a.align',\n",
              " 'data/alignments/s1/bgia2n.align',\n",
              " 'data/alignments/s1/bgia3s.align',\n",
              " 'data/alignments/s1/bgia4p.align',\n",
              " 'data/alignments/s1/bgia5a.align',\n",
              " 'data/alignments/s1/bgig6n.align',\n",
              " 'data/alignments/s1/bgig7s.align',\n",
              " 'data/alignments/s1/bgig8p.align',\n",
              " 'data/alignments/s1/bgig9a.align',\n",
              " 'data/alignments/s1/bgin1s.align',\n",
              " 'data/alignments/s1/bgin2p.align',\n",
              " 'data/alignments/s1/bgin3a.align',\n",
              " 'data/alignments/s1/bginzn.align',\n",
              " 'data/alignments/s1/bgit4n.align',\n",
              " 'data/alignments/s1/bgit5s.align',\n",
              " 'data/alignments/s1/bgit6p.align',\n",
              " 'data/alignments/s1/bgit7a.align',\n",
              " 'data/alignments/s1/bgwb4n.align',\n",
              " 'data/alignments/s1/bgwb5s.align',\n",
              " 'data/alignments/s1/bgwb6p.align',\n",
              " 'data/alignments/s1/bgwb7a.align',\n",
              " 'data/alignments/s1/bgwh8n.align',\n",
              " 'data/alignments/s1/bgwh9s.align',\n",
              " 'data/alignments/s1/bgwi1a.align',\n",
              " 'data/alignments/s1/bgwizp.align',\n",
              " 'data/alignments/s1/bgwo2n.align',\n",
              " 'data/alignments/s1/bgwo3s.align',\n",
              " 'data/alignments/s1/bgwo4p.align',\n",
              " 'data/alignments/s1/bgwo5a.align',\n",
              " 'data/alignments/s1/bgwu6n.align',\n",
              " 'data/alignments/s1/bgwu7s.align',\n",
              " 'data/alignments/s1/bgwu8p.align',\n",
              " 'data/alignments/s1/bgwu9a.align',\n",
              " 'data/alignments/s1/braf8n.align',\n",
              " 'data/alignments/s1/braf9s.align',\n",
              " 'data/alignments/s1/brag1a.align',\n",
              " 'data/alignments/s1/bragzp.align',\n",
              " 'data/alignments/s1/bram2n.align',\n",
              " 'data/alignments/s1/bram3s.align',\n",
              " 'data/alignments/s1/bram4p.align',\n",
              " 'data/alignments/s1/bram5a.align',\n",
              " 'data/alignments/s1/bras6n.align',\n",
              " 'data/alignments/s1/bras7s.align',\n",
              " 'data/alignments/s1/bras8p.align',\n",
              " 'data/alignments/s1/bras9a.align',\n",
              " 'data/alignments/s1/brba1a.align',\n",
              " 'data/alignments/s1/brbazp.align',\n",
              " 'data/alignments/s1/brbg2n.align',\n",
              " 'data/alignments/s1/brbg3s.align',\n",
              " 'data/alignments/s1/brbg4p.align',\n",
              " 'data/alignments/s1/brbg5a.align',\n",
              " 'data/alignments/s1/brbm6n.align',\n",
              " 'data/alignments/s1/brbm7s.align',\n",
              " 'data/alignments/s1/brbm8p.align',\n",
              " 'data/alignments/s1/brbm9a.align',\n",
              " 'data/alignments/s1/brbt1s.align',\n",
              " 'data/alignments/s1/brbt2p.align',\n",
              " 'data/alignments/s1/brbt3a.align',\n",
              " 'data/alignments/s1/brbtzn.align',\n",
              " 'data/alignments/s1/brif4n.align',\n",
              " 'data/alignments/s1/brif5s.align',\n",
              " 'data/alignments/s1/brif6p.align',\n",
              " 'data/alignments/s1/brif7a.align',\n",
              " 'data/alignments/s1/bril8n.align',\n",
              " 'data/alignments/s1/bril9s.align',\n",
              " 'data/alignments/s1/brim1a.align',\n",
              " 'data/alignments/s1/brimzp.align',\n",
              " 'data/alignments/s1/bris2n.align',\n",
              " 'data/alignments/s1/bris3s.align',\n",
              " 'data/alignments/s1/bris4p.align',\n",
              " 'data/alignments/s1/bris5a.align',\n",
              " 'data/alignments/s1/briz6n.align',\n",
              " 'data/alignments/s1/briz7s.align',\n",
              " 'data/alignments/s1/briz8p.align',\n",
              " 'data/alignments/s1/briz9a.align',\n",
              " 'data/alignments/s1/brwa2n.align',\n",
              " 'data/alignments/s1/brwa3s.align',\n",
              " 'data/alignments/s1/brwa4p.align',\n",
              " 'data/alignments/s1/brwa5a.align',\n",
              " 'data/alignments/s1/brwg6n.align',\n",
              " 'data/alignments/s1/brwg7s.align',\n",
              " 'data/alignments/s1/brwg8p.align',\n",
              " 'data/alignments/s1/brwg9a.align',\n",
              " 'data/alignments/s1/brwn1s.align',\n",
              " 'data/alignments/s1/brwn2p.align',\n",
              " 'data/alignments/s1/brwn3a.align',\n",
              " 'data/alignments/s1/brwnzn.align',\n",
              " 'data/alignments/s1/brwt4n.align',\n",
              " 'data/alignments/s1/brwt5s.align',\n",
              " 'data/alignments/s1/brwt6p.align',\n",
              " 'data/alignments/s1/brwt7a.align',\n",
              " 'data/alignments/s1/bwaa1s.align',\n",
              " 'data/alignments/s1/bwaa2p.align',\n",
              " 'data/alignments/s1/bwaa3a.align',\n",
              " 'data/alignments/s1/bwaazn.align',\n",
              " 'data/alignments/s1/bwag4n.align',\n",
              " 'data/alignments/s1/bwag5s.align',\n",
              " 'data/alignments/s1/bwag6p.align',\n",
              " 'data/alignments/s1/bwag7a.align',\n",
              " 'data/alignments/s1/bwam8n.align',\n",
              " 'data/alignments/s1/bwam9s.align',\n",
              " 'data/alignments/s1/bwan1a.align',\n",
              " 'data/alignments/s1/bwanzp.align',\n",
              " 'data/alignments/s1/bwat2n.align',\n",
              " 'data/alignments/s1/bwat3s.align',\n",
              " 'data/alignments/s1/bwat4p.align',\n",
              " 'data/alignments/s1/bwat5a.align',\n",
              " 'data/alignments/s1/bwba4n.align',\n",
              " 'data/alignments/s1/bwba5s.align',\n",
              " 'data/alignments/s1/bwba6p.align',\n",
              " 'data/alignments/s1/bwba7a.align',\n",
              " 'data/alignments/s1/bwbg8n.align',\n",
              " 'data/alignments/s1/bwbg9s.align',\n",
              " 'data/alignments/s1/bwbh1a.align',\n",
              " 'data/alignments/s1/bwbhzp.align',\n",
              " 'data/alignments/s1/bwbn2n.align',\n",
              " 'data/alignments/s1/bwbn3s.align',\n",
              " 'data/alignments/s1/bwbn4p.align',\n",
              " 'data/alignments/s1/bwbn5a.align',\n",
              " 'data/alignments/s1/bwbt6n.align',\n",
              " 'data/alignments/s1/bwbt7s.align',\n",
              " 'data/alignments/s1/bwbt8p.align',\n",
              " 'data/alignments/s1/bwbt9a.align',\n",
              " 'data/alignments/s1/bwig1s.align',\n",
              " 'data/alignments/s1/bwig2p.align',\n",
              " 'data/alignments/s1/bwig3a.align',\n",
              " 'data/alignments/s1/bwigzn.align',\n",
              " 'data/alignments/s1/bwim4n.align',\n",
              " 'data/alignments/s1/bwim5s.align',\n",
              " 'data/alignments/s1/bwim6p.align',\n",
              " 'data/alignments/s1/bwim7a.align',\n",
              " 'data/alignments/s1/bwis8n.align',\n",
              " 'data/alignments/s1/bwis9s.align',\n",
              " 'data/alignments/s1/bwit1a.align',\n",
              " 'data/alignments/s1/bwitzp.align',\n",
              " 'data/alignments/s1/bwwa8n.align',\n",
              " 'data/alignments/s1/bwwa9s.align',\n",
              " 'data/alignments/s1/bwwb1a.align',\n",
              " 'data/alignments/s1/bwwbzp.align',\n",
              " 'data/alignments/s1/bwwh2n.align',\n",
              " 'data/alignments/s1/bwwh3s.align',\n",
              " 'data/alignments/s1/bwwh4p.align',\n",
              " 'data/alignments/s1/bwwh5a.align',\n",
              " 'data/alignments/s1/bwwn6n.align',\n",
              " 'data/alignments/s1/bwwn7s.align',\n",
              " 'data/alignments/s1/bwwn8p.align',\n",
              " 'data/alignments/s1/bwwn9a.align',\n",
              " 'data/alignments/s1/bwwu1s.align',\n",
              " 'data/alignments/s1/bwwu2p.align',\n",
              " 'data/alignments/s1/bwwu3a.align',\n",
              " 'data/alignments/s1/bwwuzn.align',\n",
              " 'data/alignments/s1/lbad6n.align',\n",
              " 'data/alignments/s1/lbad7s.align',\n",
              " 'data/alignments/s1/lbad8p.align',\n",
              " 'data/alignments/s1/lbad9a.align',\n",
              " 'data/alignments/s1/lbak1s.align',\n",
              " 'data/alignments/s1/lbak2p.align',\n",
              " 'data/alignments/s1/lbak3a.align',\n",
              " 'data/alignments/s1/lbakzn.align',\n",
              " 'data/alignments/s1/lbaq4n.align',\n",
              " 'data/alignments/s1/lbaq5s.align',\n",
              " 'data/alignments/s1/lbaq6p.align',\n",
              " 'data/alignments/s1/lbaq7a.align',\n",
              " 'data/alignments/s1/lbax8n.align',\n",
              " 'data/alignments/s1/lbax9s.align',\n",
              " 'data/alignments/s1/lbay1a.align',\n",
              " 'data/alignments/s1/lbayzp.align',\n",
              " 'data/alignments/s1/lbbe1s.align',\n",
              " 'data/alignments/s1/lbbe2p.align',\n",
              " 'data/alignments/s1/lbbe3a.align',\n",
              " 'data/alignments/s1/lbbezn.align',\n",
              " 'data/alignments/s1/lbbk4n.align',\n",
              " 'data/alignments/s1/lbbk5s.align',\n",
              " 'data/alignments/s1/lbbk6p.align',\n",
              " 'data/alignments/s1/lbbk7a.align',\n",
              " 'data/alignments/s1/lbbq8n.align',\n",
              " 'data/alignments/s1/lbbq9s.align',\n",
              " 'data/alignments/s1/lbbr1a.align',\n",
              " 'data/alignments/s1/lbbrzp.align',\n",
              " 'data/alignments/s1/lbby2n.align',\n",
              " 'data/alignments/s1/lbby3s.align',\n",
              " 'data/alignments/s1/lbby4p.align',\n",
              " 'data/alignments/s1/lbby5a.align',\n",
              " 'data/alignments/s1/lbid2n.align',\n",
              " 'data/alignments/s1/lbid3s.align',\n",
              " 'data/alignments/s1/lbid4p.align',\n",
              " 'data/alignments/s1/lbid5a.align',\n",
              " 'data/alignments/s1/lbij6n.align',\n",
              " 'data/alignments/s1/lbij7s.align',\n",
              " 'data/alignments/s1/lbij8p.align',\n",
              " 'data/alignments/s1/lbij9a.align',\n",
              " 'data/alignments/s1/lbiq1s.align',\n",
              " 'data/alignments/s1/lbiq2p.align',\n",
              " 'data/alignments/s1/lbiq3a.align',\n",
              " 'data/alignments/s1/lbiqzn.align',\n",
              " 'data/alignments/s1/lbix4n.align',\n",
              " 'data/alignments/s1/lbix5s.align',\n",
              " 'data/alignments/s1/lbix6p.align',\n",
              " 'data/alignments/s1/lbix7a.align',\n",
              " 'data/alignments/s1/lbwe4n.align',\n",
              " 'data/alignments/s1/lbwe5s.align',\n",
              " 'data/alignments/s1/lbwe6p.align',\n",
              " 'data/alignments/s1/lbwe7a.align',\n",
              " 'data/alignments/s1/lbwk8n.align',\n",
              " 'data/alignments/s1/lbwk9s.align',\n",
              " 'data/alignments/s1/lbwl1a.align',\n",
              " 'data/alignments/s1/lbwlzp.align',\n",
              " 'data/alignments/s1/lbwr2n.align',\n",
              " 'data/alignments/s1/lbwr3s.align',\n",
              " 'data/alignments/s1/lbwr4p.align',\n",
              " 'data/alignments/s1/lbwr5a.align',\n",
              " 'data/alignments/s1/lbwy6n.align',\n",
              " 'data/alignments/s1/lbwy7s.align',\n",
              " 'data/alignments/s1/lbwy8p.align',\n",
              " 'data/alignments/s1/lbwy9a.align',\n",
              " 'data/alignments/s1/lgaf4n.align',\n",
              " 'data/alignments/s1/lgaf5s.align',\n",
              " 'data/alignments/s1/lgaf6p.align',\n",
              " 'data/alignments/s1/lgaf7a.align',\n",
              " 'data/alignments/s1/lgal8n.align',\n",
              " 'data/alignments/s1/lgal9s.align',\n",
              " 'data/alignments/s1/lgam1a.align',\n",
              " 'data/alignments/s1/lgamzp.align',\n",
              " 'data/alignments/s1/lgas2n.align',\n",
              " 'data/alignments/s1/lgas3s.align',\n",
              " 'data/alignments/s1/lgas4p.align',\n",
              " 'data/alignments/s1/lgas5a.align',\n",
              " 'data/alignments/s1/lgaz6n.align',\n",
              " 'data/alignments/s1/lgaz7s.align',\n",
              " 'data/alignments/s1/lgaz8p.align',\n",
              " 'data/alignments/s1/lgaz9a.align',\n",
              " 'data/alignments/s1/lgbf8n.align',\n",
              " 'data/alignments/s1/lgbf9s.align',\n",
              " 'data/alignments/s1/lgbg1a.align',\n",
              " 'data/alignments/s1/lgbgzp.align',\n",
              " 'data/alignments/s1/lgbm2n.align',\n",
              " 'data/alignments/s1/lgbm3s.align',\n",
              " 'data/alignments/s1/lgbm4p.align',\n",
              " 'data/alignments/s1/lgbm5a.align',\n",
              " 'data/alignments/s1/lgbs6n.align',\n",
              " 'data/alignments/s1/lgbs7s.align',\n",
              " 'data/alignments/s1/lgbs8p.align',\n",
              " 'data/alignments/s1/lgbs9a.align',\n",
              " 'data/alignments/s1/lgif1s.align',\n",
              " 'data/alignments/s1/lgif2p.align',\n",
              " 'data/alignments/s1/lgif3a.align',\n",
              " 'data/alignments/s1/lgifzn.align',\n",
              " 'data/alignments/s1/lgil4n.align',\n",
              " 'data/alignments/s1/lgil5s.align',\n",
              " 'data/alignments/s1/lgil6p.align',\n",
              " 'data/alignments/s1/lgil7a.align',\n",
              " 'data/alignments/s1/lgir8n.align',\n",
              " 'data/alignments/s1/lgir9s.align',\n",
              " 'data/alignments/s1/lgis1a.align',\n",
              " 'data/alignments/s1/lgiszp.align',\n",
              " 'data/alignments/s1/lgiz2n.align',\n",
              " 'data/alignments/s1/lgiz3s.align',\n",
              " 'data/alignments/s1/lgiz4p.align',\n",
              " 'data/alignments/s1/lgiz5a.align',\n",
              " 'data/alignments/s1/lgwa1a.align',\n",
              " 'data/alignments/s1/lgwazp.align',\n",
              " 'data/alignments/s1/lgwg2n.align',\n",
              " 'data/alignments/s1/lgwg3s.align',\n",
              " 'data/alignments/s1/lgwg4p.align',\n",
              " 'data/alignments/s1/lgwg5a.align',\n",
              " 'data/alignments/s1/lgwm6n.align',\n",
              " 'data/alignments/s1/lgwm7s.align',\n",
              " 'data/alignments/s1/lgwm8p.align',\n",
              " 'data/alignments/s1/lgwm9a.align',\n",
              " 'data/alignments/s1/lgwt1s.align',\n",
              " 'data/alignments/s1/lgwt2p.align',\n",
              " 'data/alignments/s1/lgwt3a.align',\n",
              " 'data/alignments/s1/lgwtzn.align',\n",
              " 'data/alignments/s1/lrae2n.align',\n",
              " 'data/alignments/s1/lrae3s.align',\n",
              " 'data/alignments/s1/lrae4p.align',\n",
              " 'data/alignments/s1/lrae5a.align',\n",
              " 'data/alignments/s1/lrak6n.align',\n",
              " 'data/alignments/s1/lrak7s.align',\n",
              " 'data/alignments/s1/lrak8p.align',\n",
              " 'data/alignments/s1/lrak9a.align',\n",
              " 'data/alignments/s1/lrar1s.align',\n",
              " 'data/alignments/s1/lrar2p.align',\n",
              " 'data/alignments/s1/lrar3a.align',\n",
              " 'data/alignments/s1/lrarzn.align',\n",
              " 'data/alignments/s1/lray4n.align',\n",
              " 'data/alignments/s1/lray5s.align',\n",
              " 'data/alignments/s1/lray6p.align',\n",
              " 'data/alignments/s1/lray7a.align',\n",
              " 'data/alignments/s1/lrbe6n.align',\n",
              " 'data/alignments/s1/lrbe7s.align',\n",
              " 'data/alignments/s1/lrbe8p.align',\n",
              " 'data/alignments/s1/lrbe9a.align',\n",
              " 'data/alignments/s1/lrbl1s.align',\n",
              " 'data/alignments/s1/lrbl2p.align',\n",
              " 'data/alignments/s1/lrbl3a.align',\n",
              " 'data/alignments/s1/lrblzn.align',\n",
              " 'data/alignments/s1/lrbr4n.align',\n",
              " 'data/alignments/s1/lrbr5s.align',\n",
              " 'data/alignments/s1/lrbr6p.align',\n",
              " 'data/alignments/s1/lrbr7a.align',\n",
              " 'data/alignments/s1/lrby8n.align',\n",
              " 'data/alignments/s1/lrby9s.align',\n",
              " 'data/alignments/s1/lrbz1a.align',\n",
              " 'data/alignments/s1/lrbzzp.align',\n",
              " 'data/alignments/s1/lrid8n.align',\n",
              " 'data/alignments/s1/lrid9s.align',\n",
              " 'data/alignments/s1/lrie1a.align',\n",
              " 'data/alignments/s1/lriezp.align',\n",
              " 'data/alignments/s1/lrik2n.align',\n",
              " 'data/alignments/s1/lrik3s.align',\n",
              " 'data/alignments/s1/lrik4p.align',\n",
              " 'data/alignments/s1/lrik5a.align',\n",
              " 'data/alignments/s1/lriq6n.align',\n",
              " 'data/alignments/s1/lriq7s.align',\n",
              " 'data/alignments/s1/lriq8p.align',\n",
              " 'data/alignments/s1/lriq9a.align',\n",
              " 'data/alignments/s1/lriy1s.align',\n",
              " 'data/alignments/s1/lriy2p.align',\n",
              " 'data/alignments/s1/lriy3a.align',\n",
              " 'data/alignments/s1/lriyzn.align',\n",
              " 'data/alignments/s1/lrwf1s.align',\n",
              " 'data/alignments/s1/lrwf2p.align',\n",
              " 'data/alignments/s1/lrwf3a.align',\n",
              " 'data/alignments/s1/lrwfzn.align',\n",
              " 'data/alignments/s1/lrwl4n.align',\n",
              " 'data/alignments/s1/lrwl5s.align',\n",
              " 'data/alignments/s1/lrwl6p.align',\n",
              " 'data/alignments/s1/lrwl7a.align',\n",
              " 'data/alignments/s1/lrwr8n.align',\n",
              " 'data/alignments/s1/lrwr9s.align',\n",
              " 'data/alignments/s1/lrws1a.align',\n",
              " 'data/alignments/s1/lrwszp.align',\n",
              " 'data/alignments/s1/lrwz2n.align',\n",
              " 'data/alignments/s1/lrwz3s.align',\n",
              " 'data/alignments/s1/lrwz4p.align',\n",
              " 'data/alignments/s1/lrwz5a.align',\n",
              " 'data/alignments/s1/lwae8n.align',\n",
              " 'data/alignments/s1/lwae9s.align',\n",
              " 'data/alignments/s1/lwaf1a.align',\n",
              " 'data/alignments/s1/lwafzp.align',\n",
              " 'data/alignments/s1/lwal2n.align',\n",
              " 'data/alignments/s1/lwal3s.align',\n",
              " 'data/alignments/s1/lwal4p.align',\n",
              " 'data/alignments/s1/lwal5a.align',\n",
              " 'data/alignments/s1/lwar6n.align',\n",
              " 'data/alignments/s1/lwar7s.align',\n",
              " 'data/alignments/s1/lwar8p.align',\n",
              " 'data/alignments/s1/lwar9a.align',\n",
              " 'data/alignments/s1/lwaz1s.align',\n",
              " 'data/alignments/s1/lwaz2p.align',\n",
              " 'data/alignments/s1/lwaz3a.align',\n",
              " 'data/alignments/s1/lwazzn.align',\n",
              " 'data/alignments/s1/lwbf2n.align',\n",
              " 'data/alignments/s1/lwbf3s.align',\n",
              " 'data/alignments/s1/lwbf4p.align',\n",
              " 'data/alignments/s1/lwbf5a.align',\n",
              " 'data/alignments/s1/lwbl6n.align',\n",
              " 'data/alignments/s1/lwbl7s.align',\n",
              " 'data/alignments/s1/lwbl8p.align',\n",
              " 'data/alignments/s1/lwbl9a.align',\n",
              " 'data/alignments/s1/lwbs1s.align',\n",
              " 'data/alignments/s1/lwbs2p.align',\n",
              " 'data/alignments/s1/lwbs3a.align',\n",
              " 'data/alignments/s1/lwbszn.align',\n",
              " 'data/alignments/s1/lwbz4n.align',\n",
              " 'data/alignments/s1/lwbz5s.align',\n",
              " 'data/alignments/s1/lwbz6p.align',\n",
              " 'data/alignments/s1/lwbz7a.align',\n",
              " 'data/alignments/s1/lwie4n.align',\n",
              " 'data/alignments/s1/lwie5s.align',\n",
              " 'data/alignments/s1/lwie6p.align',\n",
              " 'data/alignments/s1/lwie7a.align',\n",
              " 'data/alignments/s1/lwik8n.align',\n",
              " 'data/alignments/s1/lwik9s.align',\n",
              " 'data/alignments/s1/lwil1a.align',\n",
              " 'data/alignments/s1/lwilzp.align',\n",
              " 'data/alignments/s1/lwir2n.align',\n",
              " 'data/alignments/s1/lwir3s.align',\n",
              " 'data/alignments/s1/lwir4p.align',\n",
              " 'data/alignments/s1/lwir5a.align',\n",
              " 'data/alignments/s1/lwiy6n.align',\n",
              " 'data/alignments/s1/lwiy7s.align',\n",
              " 'data/alignments/s1/lwiy8p.align',\n",
              " 'data/alignments/s1/lwiy9a.align',\n",
              " 'data/alignments/s1/lwwf6n.align',\n",
              " 'data/alignments/s1/lwwf7s.align',\n",
              " 'data/alignments/s1/lwwf8p.align',\n",
              " 'data/alignments/s1/lwwf9a.align',\n",
              " 'data/alignments/s1/lwwm1s.align',\n",
              " 'data/alignments/s1/lwwm2p.align',\n",
              " 'data/alignments/s1/lwwm3a.align',\n",
              " 'data/alignments/s1/lwwmzn.align',\n",
              " 'data/alignments/s1/lwws4n.align',\n",
              " 'data/alignments/s1/lwws5s.align',\n",
              " 'data/alignments/s1/lwws6p.align',\n",
              " 'data/alignments/s1/lwws7a.align',\n",
              " 'data/alignments/s1/lwwz8n.align',\n",
              " 'data/alignments/s1/lwwz9s.align',\n",
              " 'data/alignments/s1/pbac1s.align',\n",
              " 'data/alignments/s1/pbac2p.align',\n",
              " 'data/alignments/s1/pbac3a.align',\n",
              " 'data/alignments/s1/pbaczn.align',\n",
              " 'data/alignments/s1/pbai4n.align',\n",
              " 'data/alignments/s1/pbai5s.align',\n",
              " 'data/alignments/s1/pbai6p.align',\n",
              " 'data/alignments/s1/pbai7a.align',\n",
              " 'data/alignments/s1/pbao8n.align',\n",
              " 'data/alignments/s1/pbao9s.align',\n",
              " 'data/alignments/s1/pbap1a.align',\n",
              " 'data/alignments/s1/pbapzp.align',\n",
              " 'data/alignments/s1/pbav2n.align',\n",
              " 'data/alignments/s1/pbav3s.align',\n",
              " 'data/alignments/s1/pbav4p.align',\n",
              " 'data/alignments/s1/pbav5a.align',\n",
              " 'data/alignments/s1/pbbc4n.align',\n",
              " 'data/alignments/s1/pbbc5s.align',\n",
              " 'data/alignments/s1/pbbc6p.align',\n",
              " 'data/alignments/s1/pbbc7a.align',\n",
              " 'data/alignments/s1/pbbi8n.align',\n",
              " 'data/alignments/s1/pbbi9s.align',\n",
              " 'data/alignments/s1/pbbj1a.align',\n",
              " 'data/alignments/s1/pbbjzp.align',\n",
              " 'data/alignments/s1/pbbp2n.align',\n",
              " 'data/alignments/s1/pbbp3s.align',\n",
              " 'data/alignments/s1/pbbp4p.align',\n",
              " 'data/alignments/s1/pbbp5a.align',\n",
              " 'data/alignments/s1/pbbv6n.align',\n",
              " 'data/alignments/s1/pbbv7s.align',\n",
              " 'data/alignments/s1/pbbv8p.align',\n",
              " 'data/alignments/s1/pbbv9a.align',\n",
              " 'data/alignments/s1/pbib6n.align',\n",
              " 'data/alignments/s1/pbib7s.align',\n",
              " 'data/alignments/s1/pbib8p.align',\n",
              " 'data/alignments/s1/pbib9a.align',\n",
              " 'data/alignments/s1/pbii1s.align',\n",
              " 'data/alignments/s1/pbii2p.align',\n",
              " 'data/alignments/s1/pbii3a.align',\n",
              " 'data/alignments/s1/pbiizn.align',\n",
              " 'data/alignments/s1/pbio4n.align',\n",
              " 'data/alignments/s1/pbio5s.align',\n",
              " 'data/alignments/s1/pbio6p.align',\n",
              " 'data/alignments/s1/pbio7a.align',\n",
              " 'data/alignments/s1/pbiu8n.align',\n",
              " 'data/alignments/s1/pbiu9s.align',\n",
              " 'data/alignments/s1/pbiv1a.align',\n",
              " 'data/alignments/s1/pbivzp.align',\n",
              " 'data/alignments/s1/pbwc8n.align',\n",
              " 'data/alignments/s1/pbwc9s.align',\n",
              " 'data/alignments/s1/pbwd1a.align',\n",
              " 'data/alignments/s1/pbwdzp.align',\n",
              " 'data/alignments/s1/pbwj2n.align',\n",
              " 'data/alignments/s1/pbwj3s.align',\n",
              " 'data/alignments/s1/pbwj4p.align',\n",
              " 'data/alignments/s1/pbwj5a.align',\n",
              " 'data/alignments/s1/pbwp6n.align',\n",
              " 'data/alignments/s1/pbwp7s.align',\n",
              " 'data/alignments/s1/pbwp8p.align',\n",
              " 'data/alignments/s1/pbwp9a.align',\n",
              " 'data/alignments/s1/pbwx1s.align',\n",
              " 'data/alignments/s1/pbwx2p.align',\n",
              " 'data/alignments/s1/pbwx3a.align',\n",
              " 'data/alignments/s1/pbwxzn.align',\n",
              " 'data/alignments/s1/pgad8n.align',\n",
              " 'data/alignments/s1/pgad9s.align',\n",
              " 'data/alignments/s1/pgae1a.align',\n",
              " 'data/alignments/s1/pgaezp.align',\n",
              " 'data/alignments/s1/pgak2n.align',\n",
              " 'data/alignments/s1/pgak3s.align',\n",
              " 'data/alignments/s1/pgak4p.align',\n",
              " 'data/alignments/s1/pgak5a.align',\n",
              " 'data/alignments/s1/pgaq6n.align',\n",
              " 'data/alignments/s1/pgaq7s.align',\n",
              " 'data/alignments/s1/pgaq8p.align',\n",
              " 'data/alignments/s1/pgaq9a.align',\n",
              " 'data/alignments/s1/pgay1s.align',\n",
              " 'data/alignments/s1/pgay2p.align',\n",
              " 'data/alignments/s1/pgay3a.align',\n",
              " 'data/alignments/s1/pgayzn.align',\n",
              " 'data/alignments/s1/pgbe2n.align',\n",
              " 'data/alignments/s1/pgbe3s.align',\n",
              " 'data/alignments/s1/pgbe4p.align',\n",
              " 'data/alignments/s1/pgbe5a.align',\n",
              " 'data/alignments/s1/pgbk6n.align',\n",
              " 'data/alignments/s1/pgbk7s.align',\n",
              " 'data/alignments/s1/pgbk8p.align',\n",
              " 'data/alignments/s1/pgbk9a.align',\n",
              " 'data/alignments/s1/pgbr1s.align',\n",
              " 'data/alignments/s1/pgbr2p.align',\n",
              " 'data/alignments/s1/pgbr3a.align',\n",
              " 'data/alignments/s1/pgbrzn.align',\n",
              " 'data/alignments/s1/pgby4n.align',\n",
              " 'data/alignments/s1/pgby5s.align',\n",
              " 'data/alignments/s1/pgby6p.align',\n",
              " 'data/alignments/s1/pgby7a.align',\n",
              " 'data/alignments/s1/pgid4n.align',\n",
              " 'data/alignments/s1/pgid5s.align',\n",
              " 'data/alignments/s1/pgid6p.align',\n",
              " 'data/alignments/s1/pgid7a.align',\n",
              " 'data/alignments/s1/pgij8n.align',\n",
              " 'data/alignments/s1/pgij9s.align',\n",
              " 'data/alignments/s1/pgik1a.align',\n",
              " 'data/alignments/s1/pgikzp.align',\n",
              " 'data/alignments/s1/pgiq2n.align',\n",
              " 'data/alignments/s1/pgiq3s.align',\n",
              " 'data/alignments/s1/pgiq4p.align',\n",
              " 'data/alignments/s1/pgiq5a.align',\n",
              " 'data/alignments/s1/pgix6n.align',\n",
              " 'data/alignments/s1/pgix7s.align',\n",
              " 'data/alignments/s1/pgix8p.align',\n",
              " 'data/alignments/s1/pgix9a.align',\n",
              " 'data/alignments/s1/pgwe6n.align',\n",
              " 'data/alignments/s1/pgwe7s.align',\n",
              " 'data/alignments/s1/pgwe8p.align',\n",
              " 'data/alignments/s1/pgwe9a.align',\n",
              " 'data/alignments/s1/pgwl1s.align',\n",
              " 'data/alignments/s1/pgwl2p.align',\n",
              " 'data/alignments/s1/pgwl3a.align',\n",
              " 'data/alignments/s1/pgwlzn.align',\n",
              " 'data/alignments/s1/pgwr4n.align',\n",
              " 'data/alignments/s1/pgwr5s.align',\n",
              " 'data/alignments/s1/pgwr6p.align',\n",
              " 'data/alignments/s1/pgwr7a.align',\n",
              " 'data/alignments/s1/pgwy8n.align',\n",
              " 'data/alignments/s1/pgwy9s.align',\n",
              " 'data/alignments/s1/pgwz1a.align',\n",
              " 'data/alignments/s1/pgwzzp.align',\n",
              " 'data/alignments/s1/prac6n.align',\n",
              " 'data/alignments/s1/prac7s.align',\n",
              " 'data/alignments/s1/prac8p.align',\n",
              " 'data/alignments/s1/prac9a.align',\n",
              " 'data/alignments/s1/praj1s.align',\n",
              " 'data/alignments/s1/praj2p.align',\n",
              " 'data/alignments/s1/praj3a.align',\n",
              " 'data/alignments/s1/prajzn.align',\n",
              " 'data/alignments/s1/prap4n.align',\n",
              " 'data/alignments/s1/prap5s.align',\n",
              " 'data/alignments/s1/prap6p.align',\n",
              " 'data/alignments/s1/prap7a.align',\n",
              " 'data/alignments/s1/prav8n.align',\n",
              " 'data/alignments/s1/prav9s.align',\n",
              " 'data/alignments/s1/prax1a.align',\n",
              " 'data/alignments/s1/praxzp.align',\n",
              " 'data/alignments/s1/prbd1s.align',\n",
              " 'data/alignments/s1/prbd2p.align',\n",
              " 'data/alignments/s1/prbd3a.align',\n",
              " 'data/alignments/s1/prbdzn.align',\n",
              " 'data/alignments/s1/prbj4n.align',\n",
              " 'data/alignments/s1/prbj5s.align',\n",
              " 'data/alignments/s1/prbj6p.align',\n",
              " 'data/alignments/s1/prbj7a.align',\n",
              " 'data/alignments/s1/prbp8n.align',\n",
              " 'data/alignments/s1/prbp9s.align',\n",
              " 'data/alignments/s1/prbq1a.align',\n",
              " 'data/alignments/s1/prbqzp.align',\n",
              " 'data/alignments/s1/prbx2n.align',\n",
              " 'data/alignments/s1/prbx3s.align',\n",
              " 'data/alignments/s1/prbx4p.align',\n",
              " 'data/alignments/s1/prbx5a.align',\n",
              " 'data/alignments/s1/pric2n.align',\n",
              " 'data/alignments/s1/pric3s.align',\n",
              " 'data/alignments/s1/pric4p.align',\n",
              " 'data/alignments/s1/pric5a.align',\n",
              " 'data/alignments/s1/prii6n.align',\n",
              " 'data/alignments/s1/prii7s.align',\n",
              " 'data/alignments/s1/prii8p.align',\n",
              " 'data/alignments/s1/prii9a.align',\n",
              " 'data/alignments/s1/prip1s.align',\n",
              " 'data/alignments/s1/prip2p.align',\n",
              " 'data/alignments/s1/prip3a.align',\n",
              " 'data/alignments/s1/pripzn.align',\n",
              " 'data/alignments/s1/priv4n.align',\n",
              " 'data/alignments/s1/priv5s.align',\n",
              " 'data/alignments/s1/priv6p.align',\n",
              " 'data/alignments/s1/priv7a.align',\n",
              " 'data/alignments/s1/prwd4n.align',\n",
              " 'data/alignments/s1/prwd5s.align',\n",
              " 'data/alignments/s1/prwd6p.align',\n",
              " 'data/alignments/s1/prwd7a.align',\n",
              " 'data/alignments/s1/prwj8n.align',\n",
              " 'data/alignments/s1/prwj9s.align',\n",
              " 'data/alignments/s1/prwk1a.align',\n",
              " 'data/alignments/s1/prwkzp.align',\n",
              " 'data/alignments/s1/prwq2n.align',\n",
              " 'data/alignments/s1/prwq3s.align',\n",
              " 'data/alignments/s1/prwq4p.align',\n",
              " 'data/alignments/s1/prwq5a.align',\n",
              " 'data/alignments/s1/prwx6n.align',\n",
              " 'data/alignments/s1/prwx7s.align',\n",
              " 'data/alignments/s1/prwx8p.align',\n",
              " 'data/alignments/s1/prwx9a.align',\n",
              " 'data/alignments/s1/pwad2n.align',\n",
              " 'data/alignments/s1/pwad3s.align',\n",
              " 'data/alignments/s1/pwad4p.align',\n",
              " 'data/alignments/s1/pwad5a.align',\n",
              " 'data/alignments/s1/pwaj6n.align',\n",
              " 'data/alignments/s1/pwaj7s.align',\n",
              " 'data/alignments/s1/pwaj8p.align',\n",
              " 'data/alignments/s1/pwaj9a.align',\n",
              " 'data/alignments/s1/pwaq1s.align',\n",
              " 'data/alignments/s1/pwaq2p.align',\n",
              " 'data/alignments/s1/pwaq3a.align',\n",
              " 'data/alignments/s1/pwaqzn.align',\n",
              " 'data/alignments/s1/pwax4n.align',\n",
              " 'data/alignments/s1/pwax5s.align',\n",
              " 'data/alignments/s1/pwax6p.align',\n",
              " 'data/alignments/s1/pwax7a.align',\n",
              " 'data/alignments/s1/pwbd6n.align',\n",
              " 'data/alignments/s1/pwbd7s.align',\n",
              " 'data/alignments/s1/pwbd8p.align',\n",
              " 'data/alignments/s1/pwbd9a.align',\n",
              " 'data/alignments/s1/pwbk1s.align',\n",
              " 'data/alignments/s1/pwbk2p.align',\n",
              " 'data/alignments/s1/pwbk3a.align',\n",
              " 'data/alignments/s1/pwbkzn.align',\n",
              " 'data/alignments/s1/pwbq4n.align',\n",
              " 'data/alignments/s1/pwbq5s.align',\n",
              " 'data/alignments/s1/pwbq6p.align',\n",
              " 'data/alignments/s1/pwbq7a.align',\n",
              " 'data/alignments/s1/pwbx8n.align',\n",
              " 'data/alignments/s1/pwbx9s.align',\n",
              " 'data/alignments/s1/pwby1a.align',\n",
              " 'data/alignments/s1/pwbyzp.align',\n",
              " 'data/alignments/s1/pwic8n.align',\n",
              " 'data/alignments/s1/pwic9s.align',\n",
              " 'data/alignments/s1/pwid1a.align',\n",
              " 'data/alignments/s1/pwidzp.align',\n",
              " 'data/alignments/s1/pwij2n.align',\n",
              " 'data/alignments/s1/pwij3s.align',\n",
              " 'data/alignments/s1/pwij4p.align',\n",
              " 'data/alignments/s1/pwij5a.align',\n",
              " 'data/alignments/s1/pwip6n.align',\n",
              " 'data/alignments/s1/pwip7s.align',\n",
              " 'data/alignments/s1/pwip8p.align',\n",
              " 'data/alignments/s1/pwip9a.align',\n",
              " 'data/alignments/s1/pwix1s.align',\n",
              " 'data/alignments/s1/pwix2p.align',\n",
              " 'data/alignments/s1/pwix3a.align',\n",
              " 'data/alignments/s1/pwixzn.align',\n",
              " 'data/alignments/s1/pwwe1s.align',\n",
              " 'data/alignments/s1/pwwe2p.align',\n",
              " 'data/alignments/s1/pwwe3a.align',\n",
              " 'data/alignments/s1/pwwezn.align',\n",
              " 'data/alignments/s1/pwwk4n.align',\n",
              " 'data/alignments/s1/pwwk5s.align',\n",
              " 'data/alignments/s1/pwwk6p.align',\n",
              " 'data/alignments/s1/pwwk7a.align',\n",
              " 'data/alignments/s1/pwwq8n.align',\n",
              " 'data/alignments/s1/pwwq9s.align',\n",
              " 'data/alignments/s1/pwwr1a.align',\n",
              " 'data/alignments/s1/pwwrzp.align',\n",
              " 'data/alignments/s1/pwwy2n.align',\n",
              " 'data/alignments/s1/pwwy3s.align',\n",
              " 'data/alignments/s1/pwwy4p.align',\n",
              " 'data/alignments/s1/pwwy5a.align',\n",
              " 'data/alignments/s1/sbaa4n.align',\n",
              " 'data/alignments/s1/sbaa5s.align',\n",
              " 'data/alignments/s1/sbaa6p.align',\n",
              " 'data/alignments/s1/sbaa7a.align',\n",
              " 'data/alignments/s1/sbag8n.align',\n",
              " 'data/alignments/s1/sbag9s.align',\n",
              " 'data/alignments/s1/sbah1a.align',\n",
              " 'data/alignments/s1/sbahzp.align',\n",
              " 'data/alignments/s1/sban2n.align',\n",
              " 'data/alignments/s1/sban3s.align',\n",
              " 'data/alignments/s1/sban4p.align',\n",
              " 'data/alignments/s1/sban5a.align',\n",
              " 'data/alignments/s1/sbat6n.align',\n",
              " 'data/alignments/s1/sbat7s.align',\n",
              " 'data/alignments/s1/sbat8p.align',\n",
              " 'data/alignments/s1/sbat9a.align',\n",
              " 'data/alignments/s1/sbba8n.align',\n",
              " 'data/alignments/s1/sbba9s.align',\n",
              " 'data/alignments/s1/sbbb1a.align',\n",
              " 'data/alignments/s1/sbbbzp.align',\n",
              " 'data/alignments/s1/sbbh2n.align',\n",
              " 'data/alignments/s1/sbbh3s.align',\n",
              " 'data/alignments/s1/sbbh4p.align',\n",
              " 'data/alignments/s1/sbbh5a.align',\n",
              " 'data/alignments/s1/sbbn6n.align',\n",
              " 'data/alignments/s1/sbbn7s.align',\n",
              " 'data/alignments/s1/sbbn8p.align',\n",
              " 'data/alignments/s1/sbbn9a.align',\n",
              " 'data/alignments/s1/sbbu1s.align',\n",
              " 'data/alignments/s1/sbbu2p.align',\n",
              " 'data/alignments/s1/sbbu3a.align',\n",
              " 'data/alignments/s1/sbbuzn.align',\n",
              " 'data/alignments/s1/sbia1s.align',\n",
              " 'data/alignments/s1/sbia2p.align',\n",
              " 'data/alignments/s1/sbia3a.align',\n",
              " 'data/alignments/s1/sbiazn.align',\n",
              " 'data/alignments/s1/sbig4n.align',\n",
              " 'data/alignments/s1/sbig5s.align',\n",
              " 'data/alignments/s1/sbig6p.align',\n",
              " 'data/alignments/s1/sbig7a.align',\n",
              " 'data/alignments/s1/sbim8n.align',\n",
              " 'data/alignments/s1/sbim9s.align',\n",
              " 'data/alignments/s1/sbin1a.align',\n",
              " 'data/alignments/s1/sbinzp.align',\n",
              " 'data/alignments/s1/sbit2n.align',\n",
              " 'data/alignments/s1/sbit3s.align',\n",
              " 'data/alignments/s1/sbit4p.align',\n",
              " 'data/alignments/s1/sbit5a.align',\n",
              " 'data/alignments/s1/sbwb2n.align',\n",
              " 'data/alignments/s1/sbwb3s.align',\n",
              " 'data/alignments/s1/sbwb4p.align',\n",
              " 'data/alignments/s1/sbwb5a.align',\n",
              " 'data/alignments/s1/sbwh6n.align',\n",
              " 'data/alignments/s1/sbwh7s.align',\n",
              " 'data/alignments/s1/sbwh8p.align',\n",
              " 'data/alignments/s1/sbwh9a.align',\n",
              " 'data/alignments/s1/sbwo1s.align',\n",
              " 'data/alignments/s1/sbwo2p.align',\n",
              " 'data/alignments/s1/sbwo3a.align',\n",
              " 'data/alignments/s1/sbwozn.align',\n",
              " 'data/alignments/s1/sbwu4n.align',\n",
              " 'data/alignments/s1/sbwu5s.align',\n",
              " 'data/alignments/s1/sbwu6p.align',\n",
              " 'data/alignments/s1/sbwu7a.align',\n",
              " 'data/alignments/s1/sgac2n.align',\n",
              " 'data/alignments/s1/sgac3s.align',\n",
              " 'data/alignments/s1/sgac4p.align',\n",
              " 'data/alignments/s1/sgac5a.align',\n",
              " 'data/alignments/s1/sgai6n.align',\n",
              " 'data/alignments/s1/sgai7s.align',\n",
              " 'data/alignments/s1/sgai8p.align',\n",
              " 'data/alignments/s1/sgai9a.align',\n",
              " 'data/alignments/s1/sgap1s.align',\n",
              " 'data/alignments/s1/sgap2p.align',\n",
              " 'data/alignments/s1/sgap3a.align',\n",
              " 'data/alignments/s1/sgapzn.align',\n",
              " 'data/alignments/s1/sgav4n.align',\n",
              " 'data/alignments/s1/sgav5s.align',\n",
              " 'data/alignments/s1/sgav6p.align',\n",
              " 'data/alignments/s1/sgav7a.align',\n",
              " 'data/alignments/s1/sgbc6n.align',\n",
              " 'data/alignments/s1/sgbc7s.align',\n",
              " 'data/alignments/s1/sgbc8p.align',\n",
              " 'data/alignments/s1/sgbc9a.align',\n",
              " 'data/alignments/s1/sgbj1s.align',\n",
              " 'data/alignments/s1/sgbj2p.align',\n",
              " 'data/alignments/s1/sgbj3a.align',\n",
              " 'data/alignments/s1/sgbjzn.align',\n",
              " 'data/alignments/s1/sgbp4n.align',\n",
              " 'data/alignments/s1/sgbp5s.align',\n",
              " 'data/alignments/s1/sgbp6p.align',\n",
              " 'data/alignments/s1/sgbp7a.align',\n",
              " 'data/alignments/s1/sgbv8n.align',\n",
              " 'data/alignments/s1/sgbv9s.align',\n",
              " 'data/alignments/s1/sgbx1a.align',\n",
              " 'data/alignments/s1/sgbxzp.align',\n",
              " 'data/alignments/s1/sgib8n.align',\n",
              " 'data/alignments/s1/sgib9s.align',\n",
              " 'data/alignments/s1/sgic1a.align',\n",
              " 'data/alignments/s1/sgiczp.align',\n",
              " 'data/alignments/s1/sgii2n.align',\n",
              " 'data/alignments/s1/sgii3s.align',\n",
              " 'data/alignments/s1/sgii4p.align',\n",
              " 'data/alignments/s1/sgii5a.align',\n",
              " 'data/alignments/s1/sgio6n.align',\n",
              " 'data/alignments/s1/sgio7s.align',\n",
              " 'data/alignments/s1/sgio8p.align',\n",
              " 'data/alignments/s1/sgio9a.align',\n",
              " 'data/alignments/s1/sgiv1s.align',\n",
              " 'data/alignments/s1/sgiv2p.align',\n",
              " 'data/alignments/s1/sgiv3a.align',\n",
              " 'data/alignments/s1/sgivzn.align',\n",
              " 'data/alignments/s1/sgwd1s.align',\n",
              " 'data/alignments/s1/sgwd2p.align',\n",
              " 'data/alignments/s1/sgwd3a.align',\n",
              " 'data/alignments/s1/sgwdzn.align',\n",
              " 'data/alignments/s1/sgwj4n.align',\n",
              " 'data/alignments/s1/sgwj5s.align',\n",
              " 'data/alignments/s1/sgwj6p.align',\n",
              " 'data/alignments/s1/sgwj7a.align',\n",
              " 'data/alignments/s1/sgwp8n.align',\n",
              " 'data/alignments/s1/sgwp9s.align',\n",
              " 'data/alignments/s1/sgwq1a.align',\n",
              " 'data/alignments/s1/sgwqzp.align',\n",
              " 'data/alignments/s1/sgwx2n.align',\n",
              " 'data/alignments/s1/sgwx3s.align',\n",
              " 'data/alignments/s1/sgwx4p.align',\n",
              " 'data/alignments/s1/sgwx5a.align',\n",
              " 'data/alignments/s1/srab1s.align',\n",
              " 'data/alignments/s1/srab2p.align',\n",
              " 'data/alignments/s1/srab3a.align',\n",
              " 'data/alignments/s1/srabzn.align',\n",
              " 'data/alignments/s1/srah4n.align',\n",
              " 'data/alignments/s1/srah5s.align',\n",
              " 'data/alignments/s1/srah6p.align',\n",
              " 'data/alignments/s1/srah7a.align',\n",
              " 'data/alignments/s1/sran8n.align',\n",
              " 'data/alignments/s1/sran9s.align',\n",
              " 'data/alignments/s1/srao1a.align',\n",
              " 'data/alignments/s1/sraozp.align',\n",
              " 'data/alignments/s1/srau2n.align',\n",
              " 'data/alignments/s1/srau3s.align',\n",
              " 'data/alignments/s1/srau4p.align',\n",
              " 'data/alignments/s1/srau5a.align',\n",
              " 'data/alignments/s1/srbb4n.align',\n",
              " 'data/alignments/s1/srbb5s.align',\n",
              " 'data/alignments/s1/srbb6p.align',\n",
              " 'data/alignments/s1/srbb7a.align',\n",
              " 'data/alignments/s1/srbh8n.align',\n",
              " 'data/alignments/s1/srbh9s.align',\n",
              " 'data/alignments/s1/srbi1a.align',\n",
              " 'data/alignments/s1/srbizp.align',\n",
              " 'data/alignments/s1/srbo2n.align',\n",
              " 'data/alignments/s1/srbo3s.align',\n",
              " 'data/alignments/s1/srbo4p.align',\n",
              " 'data/alignments/s1/srbo5a.align',\n",
              " 'data/alignments/s1/srbu6n.align',\n",
              " 'data/alignments/s1/srbu7s.align',\n",
              " 'data/alignments/s1/srbu8p.align',\n",
              " 'data/alignments/s1/srbu9a.align',\n",
              " 'data/alignments/s1/sria6n.align',\n",
              " 'data/alignments/s1/sria7s.align',\n",
              " 'data/alignments/s1/sria8p.align',\n",
              " 'data/alignments/s1/sria9a.align',\n",
              " 'data/alignments/s1/srih1s.align',\n",
              " 'data/alignments/s1/srih2p.align',\n",
              " 'data/alignments/s1/srih3a.align',\n",
              " 'data/alignments/s1/srihzn.align',\n",
              " 'data/alignments/s1/srin4n.align',\n",
              " 'data/alignments/s1/srin5s.align',\n",
              " 'data/alignments/s1/srin6p.align',\n",
              " 'data/alignments/s1/srin7a.align',\n",
              " 'data/alignments/s1/srit8n.align',\n",
              " 'data/alignments/s1/srit9s.align',\n",
              " 'data/alignments/s1/sriu1a.align',\n",
              " 'data/alignments/s1/sriuzp.align',\n",
              " 'data/alignments/s1/srwb8n.align',\n",
              " 'data/alignments/s1/srwb9s.align',\n",
              " 'data/alignments/s1/srwc1a.align',\n",
              " 'data/alignments/s1/srwczp.align',\n",
              " 'data/alignments/s1/srwi2n.align',\n",
              " 'data/alignments/s1/srwi3s.align',\n",
              " 'data/alignments/s1/srwi4p.align',\n",
              " 'data/alignments/s1/srwi5a.align',\n",
              " 'data/alignments/s1/srwo6n.align',\n",
              " 'data/alignments/s1/srwo7s.align',\n",
              " 'data/alignments/s1/srwo8p.align',\n",
              " 'data/alignments/s1/srwo9a.align',\n",
              " 'data/alignments/s1/srwv1s.align',\n",
              " 'data/alignments/s1/srwv2p.align',\n",
              " 'data/alignments/s1/srwv3a.align',\n",
              " 'data/alignments/s1/srwvzn.align',\n",
              " 'data/alignments/s1/swab6n.align',\n",
              " 'data/alignments/s1/swab7s.align',\n",
              " 'data/alignments/s1/swab8p.align',\n",
              " 'data/alignments/s1/swab9a.align',\n",
              " 'data/alignments/s1/swai1s.align',\n",
              " 'data/alignments/s1/swai2p.align',\n",
              " 'data/alignments/s1/swai3a.align',\n",
              " 'data/alignments/s1/swaizn.align',\n",
              " 'data/alignments/s1/swao4n.align',\n",
              " 'data/alignments/s1/swao5s.align',\n",
              " 'data/alignments/s1/swao6p.align',\n",
              " 'data/alignments/s1/swao7a.align',\n",
              " 'data/alignments/s1/swau8n.align',\n",
              " 'data/alignments/s1/swau9s.align',\n",
              " 'data/alignments/s1/swav1a.align',\n",
              " 'data/alignments/s1/swavzp.align',\n",
              " 'data/alignments/s1/swbc1s.align',\n",
              " 'data/alignments/s1/swbc2p.align',\n",
              " 'data/alignments/s1/swbc3a.align',\n",
              " 'data/alignments/s1/swbczn.align',\n",
              " 'data/alignments/s1/swbi4n.align',\n",
              " 'data/alignments/s1/swbi5s.align',\n",
              " 'data/alignments/s1/swbi6p.align',\n",
              " 'data/alignments/s1/swbi7a.align',\n",
              " 'data/alignments/s1/swbo8n.align',\n",
              " 'data/alignments/s1/swbo9s.align',\n",
              " 'data/alignments/s1/swbp1a.align',\n",
              " 'data/alignments/s1/swbpzp.align',\n",
              " 'data/alignments/s1/swbv2n.align',\n",
              " 'data/alignments/s1/swbv3s.align',\n",
              " 'data/alignments/s1/swbv4p.align',\n",
              " 'data/alignments/s1/swbv5a.align',\n",
              " 'data/alignments/s1/swib2n.align',\n",
              " 'data/alignments/s1/swib3s.align',\n",
              " 'data/alignments/s1/swib4p.align',\n",
              " 'data/alignments/s1/swib5a.align',\n",
              " 'data/alignments/s1/swih6n.align',\n",
              " 'data/alignments/s1/swih7s.align',\n",
              " 'data/alignments/s1/swih8p.align',\n",
              " 'data/alignments/s1/swih9a.align',\n",
              " 'data/alignments/s1/swio1s.align',\n",
              " 'data/alignments/s1/swio2p.align',\n",
              " 'data/alignments/s1/swio3a.align',\n",
              " 'data/alignments/s1/swiozn.align',\n",
              " 'data/alignments/s1/swiu4n.align',\n",
              " 'data/alignments/s1/swiu5s.align',\n",
              " 'data/alignments/s1/swiu6p.align',\n",
              " 'data/alignments/s1/swiu7a.align',\n",
              " 'data/alignments/s1/swwc4n.align',\n",
              " 'data/alignments/s1/swwc5s.align',\n",
              " 'data/alignments/s1/swwc6p.align',\n",
              " 'data/alignments/s1/swwc7a.align',\n",
              " 'data/alignments/s1/swwi8n.align',\n",
              " 'data/alignments/s1/swwi9s.align',\n",
              " 'data/alignments/s1/swwj1a.align',\n",
              " 'data/alignments/s1/swwjzp.align',\n",
              " 'data/alignments/s1/swwp2n.align',\n",
              " 'data/alignments/s1/swwp3s.align',\n",
              " 'data/alignments/s1/swwp4p.align',\n",
              " 'data/alignments/s1/swwp5a.align',\n",
              " 'data/alignments/s1/swwv6n.align',\n",
              " ...]"
            ]
          },
          "metadata": {},
          "execution_count": 159
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "def load_video(path:str) -> List[float]: #takes path of video and outputs a string of floats\n",
        "\n",
        "    cap = cv2.VideoCapture(path) #create cv2 instance\n",
        "    #loop through each one of these and store the extracted frames in an array\n",
        "    frames = []\n",
        "    for _ in range(int(cap.get(cv2.CAP_PROP_FRAME_COUNT))): \n",
        "        ret, frame = cap.read()\n",
        "        frame = tf.image.rgb_to_grayscale(frame)\n",
        "        frames.append(frame[190:236,80:220,:]) #only takes the cropped lip part of each frame (static slicing)\n",
        "        #An advanced way is to use Dlib as used in the original LipNet paper.\n",
        "    cap.release()\n",
        "\n",
        "    #scaling each frame\n",
        "    mean = tf.math.reduce_mean(frames)\n",
        "    std = tf.math.reduce_std(tf.cast(frames, tf.float32))\n",
        "    return tf.cast((frames - mean), tf.float32) / std"
      ],
      "metadata": {
        "id": "42G_Vh8tEV5z"
      },
      "execution_count": 160,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#creating vocab- every single character that is expected to be encountered in our annotations\n",
        "vocab= [x for x in \"abcdefghijklmnopqrstuvwxyz'?!123456789 \"]"
      ],
      "metadata": {
        "id": "nPlCK-qyEcjp"
      },
      "execution_count": 161,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#originally from the keras ctc Automatic Speech Recognition tutorial paper\n",
        "char_to_num = tf.keras.layers.StringLookup(vocabulary=vocab, oov_token=\"\") #converts char to num\n",
        "num_to_char = tf.keras.layers.StringLookup(\n",
        "    vocabulary= char_to_num.get_vocabulary(), oov_token=\"\", invert=True) #converts num to char\n",
        "\n",
        "\n",
        "print(\n",
        "    f\"the vocabulary is : {char_to_num.get_vocabulary()}\\n\"\n",
        "    f\"size is : {char_to_num.vocabulary_size()}\"\n",
        ")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "slLq9EtUEg2n",
        "outputId": "bb911274-c44c-434a-8040-66e20df194e1"
      },
      "execution_count": 162,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "the vocabulary is : ['', 'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j', 'k', 'l', 'm', 'n', 'o', 'p', 'q', 'r', 's', 't', 'u', 'v', 'w', 'x', 'y', 'z', \"'\", '?', '!', '1', '2', '3', '4', '5', '6', '7', '8', '9', ' ']\n",
            "size is : 40\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#load up our alignments\n",
        "def load_alignments(path:str) -> List[str]: \n",
        "    with open(path, 'r') as f: \n",
        "        lines = f.readlines() \n",
        "    tokens = []\n",
        "    for line in lines:\n",
        "        line = line.split()\n",
        "        if line[2] != 'sil': \n",
        "            tokens = [*tokens,' ',line[2]]\n",
        "    return char_to_num(tf.reshape(tf.strings.unicode_split(tokens, input_encoding='UTF-8'), (-1)))[1:]\n"
      ],
      "metadata": {
        "id": "TZjVLOp4Ejxj"
      },
      "execution_count": 163,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#returning the preprocessed videos and preprocessed alignments together from their respective paths\n",
        "\n",
        "def load_data(path:str):\n",
        "  path = bytes.decode(path.numpy())\n",
        "  filename = path.split('/')[-1].split('.')[0] #colab\n",
        "  #file_name = path.split('\\\\')[-1].split('.')[0]  for windows.\n",
        "  video_path = os.path.join('data','s1',f'{filename}.mpg')\n",
        "  alignment_path = os.path.join('data','alignments','s1',f'{filename}.align')\n",
        "  frames=load_video(video_path)\n",
        "  alignments = load_alignments(alignment_path)\n",
        "\n",
        "  return frames, alignments"
      ],
      "metadata": {
        "id": "NNjJg1-SErGB"
      },
      "execution_count": 164,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "test_path = '/data/s1/bbal6n.mpg'"
      ],
      "metadata": {
        "id": "jUYL7rnGEvGB"
      },
      "execution_count": 165,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#filename\n",
        "tf.convert_to_tensor(test_path).numpy().decode('utf-8').split('\\\\')[-1].split('.')[0]"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 36
        },
        "id": "dkjwvgIXVlBQ",
        "outputId": "5ede839e-8536-4dca-feda-d90998148627"
      },
      "execution_count": 166,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "'/data/s1/bbal6n'"
            ],
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            }
          },
          "metadata": {},
          "execution_count": 166
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "frames,alignments = load_data(tf.convert_to_tensor(test_path))"
      ],
      "metadata": {
        "id": "rilRi9QQVl3L"
      },
      "execution_count": 167,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "plt.imshow(frames[0])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 244
        },
        "id": "84-yHEwsVoE9",
        "outputId": "911eab85-4e44-474e-f980-d5241137f285"
      },
      "execution_count": 168,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7ff5f03e53c0>"
            ]
          },
          "metadata": {},
          "execution_count": 168
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "plt.imshow(frames[40])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 244
        },
        "id": "VsaIcICeVqVL",
        "outputId": "8f873551-7273-45ac-a4d9-dc9731128c6e"
      },
      "execution_count": 169,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7ff5ded20f40>"
            ]
          },
          "metadata": {},
          "execution_count": 169
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "alignments"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "78aB1_xcVtVS",
        "outputId": "17a950c5-5d29-465b-af42-4b3d5e26a9b2"
      },
      "execution_count": 170,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tf.Tensor: shape=(21,), dtype=int64, numpy=\n",
              "array([ 2,  9, 14, 39,  2, 12, 21,  5, 39,  1, 20, 39, 12, 39, 19,  9, 24,\n",
              "       39, 14, 15, 23])>"
            ]
          },
          "metadata": {},
          "execution_count": 170
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "tf.strings.reduce_join([num_to_char(word) for word in alignments])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "mTMzgUx9VwyN",
        "outputId": "6ade1ebb-d34f-4451-edf7-6fac4ea23e74"
      },
      "execution_count": 171,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tf.Tensor: shape=(), dtype=string, numpy=b'bin blue at l six now'>"
            ]
          },
          "metadata": {},
          "execution_count": 171
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "def mappable_function(path:str)->List[str]:\n",
        "  result=tf.py_function(load_data, [path], (tf.float32,tf.int64))\n",
        "  return result"
      ],
      "metadata": {
        "id": "CXTRVLsVV2rI"
      },
      "execution_count": 172,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Creating Data Pipeline:"
      ],
      "metadata": {
        "id": "xZe3KiKNV5DC"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Creating our data pipeline:\n",
        "\n",
        "STEPS:\n",
        "\n",
        "1. using tensorflow data pipeline we iterate thrugh our videos path.\n",
        "2. shuffle to get a random video path everytime \n",
        "3. map the video to its alignments\n",
        "4. add padding to the alignments (length=40) and take 75 frames for each video (we dont pad each frame/image). Take 2 sets of videos & alignments in one batch.\n",
        "5. prefetching in order to optimize our pipileine to ensure we are loading and preloading while our model is training.\n",
        "\n",
        "Cool."
      ],
      "metadata": {
        "id": "qE4DXNbSV_rd"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "data = tf.data.Dataset.list_files('/content/data/s1/*.mpg')\n",
        "data = data.shuffle(500,reshuffle_each_iteration=False)\n",
        "data = data.map(mappable_function)\n",
        "data = data.padded_batch(2, padded_shapes=([75,None,None,None],[40]))\n",
        "data = data.prefetch(tf.data.AUTOTUNE)\n",
        "#Added for split\n",
        "train= data.take(450)\n",
        "test= data.skip(450)"
      ],
      "metadata": {
        "id": "QzJLimvUV4ZS"
      },
      "execution_count": 173,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "len(train)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "cIT4H9mXYZde",
        "outputId": "f116c6ad-fa9d-49c7-e25c-3ab162a27727"
      },
      "execution_count": 174,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "450"
            ]
          },
          "metadata": {},
          "execution_count": 174
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "len(test)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "mmNWkR20YRnL",
        "outputId": "3670a764-2c54-4e20-df0b-a3ce2f57f41b"
      },
      "execution_count": 175,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "50"
            ]
          },
          "metadata": {},
          "execution_count": 175
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#testing each code line in data pipeline:\n",
        "#frames,aligns = data.as_numpy_iterator().next()"
      ],
      "metadata": {
        "id": "_NmvNDsEWFHO"
      },
      "execution_count": 176,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#len(frames) #2 set of frmes in each batch."
      ],
      "metadata": {
        "id": "Jpxir0QJWHrU"
      },
      "execution_count": 177,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#plt.imshow(frames[0][29]) 29th frame of first video"
      ],
      "metadata": {
        "id": "sRGkrs6hWJQx"
      },
      "execution_count": 178,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#aligns[0] #first video alignment with padding."
      ],
      "metadata": {
        "id": "tVnmYGzBWK-U"
      },
      "execution_count": 179,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "test1= data.as_numpy_iterator()"
      ],
      "metadata": {
        "id": "cbmedRuuWMnF"
      },
      "execution_count": 180,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "val= test1.next(); val[0]"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "WtYUjB90WN7l",
        "outputId": "adfb87ba-53ad-4119-8411-16a870871b61"
      },
      "execution_count": 181,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([[[[[1.4348304 ],\n",
              "          [1.5103477 ],\n",
              "          [1.5103477 ],\n",
              "          ...,\n",
              "          [0.41534564],\n",
              "          [0.41534564],\n",
              "          [0.41534564]],\n",
              "\n",
              "         [[1.4348304 ],\n",
              "          [1.5481064 ],\n",
              "          [1.5103477 ],\n",
              "          ...,\n",
              "          [0.45310432],\n",
              "          [0.41534564],\n",
              "          [0.41534564]],\n",
              "\n",
              "         [[1.5103477 ],\n",
              "          [1.5103477 ],\n",
              "          [1.472589  ],\n",
              "          ...,\n",
              "          [0.37758693],\n",
              "          [0.30206954],\n",
              "          [0.30206954]],\n",
              "\n",
              "         ...,\n",
              "\n",
              "         [[1.0194848 ],\n",
              "          [1.0194848 ],\n",
              "          [1.0194848 ],\n",
              "          ...,\n",
              "          [0.07551739],\n",
              "          [0.07551739],\n",
              "          [0.07551739]],\n",
              "\n",
              "         [[1.0194848 ],\n",
              "          [1.0194848 ],\n",
              "          [1.0194848 ],\n",
              "          ...,\n",
              "          [0.07551739],\n",
              "          [0.07551739],\n",
              "          [0.07551739]],\n",
              "\n",
              "         [[1.0194848 ],\n",
              "          [1.0194848 ],\n",
              "          [1.0194848 ],\n",
              "          ...,\n",
              "          [0.07551739],\n",
              "          [0.07551739],\n",
              "          [0.03775869]]],\n",
              "\n",
              "\n",
              "        [[[1.3970717 ],\n",
              "          [1.5103477 ],\n",
              "          [1.4348304 ],\n",
              "          ...,\n",
              "          [0.33982825],\n",
              "          [0.30206954],\n",
              "          [0.30206954]],\n",
              "\n",
              "         [[1.5103477 ],\n",
              "          [1.5481064 ],\n",
              "          [1.5103477 ],\n",
              "          ...,\n",
              "          [0.37758693],\n",
              "          [0.30206954],\n",
              "          [0.30206954]],\n",
              "\n",
              "         [[1.4348304 ],\n",
              "          [1.4348304 ],\n",
              "          [1.4348304 ],\n",
              "          ...,\n",
              "          [0.33982825],\n",
              "          [0.37758693],\n",
              "          [0.37758693]],\n",
              "\n",
              "         ...,\n",
              "\n",
              "         [[1.0194848 ],\n",
              "          [1.0194848 ],\n",
              "          [1.0194848 ],\n",
              "          ...,\n",
              "          [0.11327608],\n",
              "          [0.11327608],\n",
              "          [0.11327608]],\n",
              "\n",
              "         [[1.0194848 ],\n",
              "          [1.0194848 ],\n",
              "          [1.0194848 ],\n",
              "          ...,\n",
              "          [0.11327608],\n",
              "          [0.11327608],\n",
              "          [0.07551739]],\n",
              "\n",
              "         [[1.0194848 ],\n",
              "          [1.0194848 ],\n",
              "          [1.0194848 ],\n",
              "          ...,\n",
              "          [0.11327608],\n",
              "          [0.07551739],\n",
              "          [0.07551739]]],\n",
              "\n",
              "\n",
              "        [[[1.3970717 ],\n",
              "          [1.472589  ],\n",
              "          [1.4348304 ],\n",
              "          ...,\n",
              "          [0.37758693],\n",
              "          [0.30206954],\n",
              "          [0.30206954]],\n",
              "\n",
              "         [[1.3970717 ],\n",
              "          [1.3970717 ],\n",
              "          [1.3215543 ],\n",
              "          ...,\n",
              "          [0.41534564],\n",
              "          [0.30206954],\n",
              "          [0.30206954]],\n",
              "\n",
              "         [[1.5103477 ],\n",
              "          [1.5103477 ],\n",
              "          [1.4348304 ],\n",
              "          ...,\n",
              "          [0.30206954],\n",
              "          [0.33982825],\n",
              "          [0.26431087]],\n",
              "\n",
              "         ...,\n",
              "\n",
              "         [[1.0194848 ],\n",
              "          [1.0194848 ],\n",
              "          [1.0194848 ],\n",
              "          ...,\n",
              "          [0.07551739],\n",
              "          [0.07551739],\n",
              "          [0.07551739]],\n",
              "\n",
              "         [[1.0194848 ],\n",
              "          [1.0194848 ],\n",
              "          [1.0194848 ],\n",
              "          ...,\n",
              "          [0.07551739],\n",
              "          [0.07551739],\n",
              "          [0.03775869]],\n",
              "\n",
              "         [[0.98172605],\n",
              "          [0.98172605],\n",
              "          [0.98172605],\n",
              "          ...,\n",
              "          [0.07551739],\n",
              "          [0.03775869],\n",
              "          [0.03775869]]],\n",
              "\n",
              "\n",
              "        ...,\n",
              "\n",
              "\n",
              "        [[[1.472589  ],\n",
              "          [1.2460369 ],\n",
              "          [1.1327608 ],\n",
              "          ...,\n",
              "          [0.33982825],\n",
              "          [0.30206954],\n",
              "          [0.30206954]],\n",
              "\n",
              "         [[1.3970717 ],\n",
              "          [1.3970717 ],\n",
              "          [0.98172605],\n",
              "          ...,\n",
              "          [0.33982825],\n",
              "          [0.30206954],\n",
              "          [0.30206954]],\n",
              "\n",
              "         [[1.359313  ],\n",
              "          [1.4348304 ],\n",
              "          [1.4348304 ],\n",
              "          ...,\n",
              "          [0.33982825],\n",
              "          [0.33982825],\n",
              "          [0.33982825]],\n",
              "\n",
              "         ...,\n",
              "\n",
              "         [[1.0572435 ],\n",
              "          [1.0572435 ],\n",
              "          [1.0572435 ],\n",
              "          ...,\n",
              "          [0.07551739],\n",
              "          [0.03775869],\n",
              "          [0.03775869]],\n",
              "\n",
              "         [[1.0572435 ],\n",
              "          [1.0572435 ],\n",
              "          [1.0572435 ],\n",
              "          ...,\n",
              "          [0.07551739],\n",
              "          [0.03775869],\n",
              "          [0.03775869]],\n",
              "\n",
              "         [[1.0572435 ],\n",
              "          [1.0572435 ],\n",
              "          [1.0572435 ],\n",
              "          ...,\n",
              "          [0.03775869],\n",
              "          [0.03775869],\n",
              "          [0.        ]]],\n",
              "\n",
              "\n",
              "        [[[1.472589  ],\n",
              "          [1.4348304 ],\n",
              "          [1.0950022 ],\n",
              "          ...,\n",
              "          [0.33982825],\n",
              "          [0.30206954],\n",
              "          [0.30206954]],\n",
              "\n",
              "         [[1.359313  ],\n",
              "          [1.5103477 ],\n",
              "          [1.0950022 ],\n",
              "          ...,\n",
              "          [0.33982825],\n",
              "          [0.30206954],\n",
              "          [0.30206954]],\n",
              "\n",
              "         [[1.4348304 ],\n",
              "          [1.3970717 ],\n",
              "          [1.4348304 ],\n",
              "          ...,\n",
              "          [0.33982825],\n",
              "          [0.33982825],\n",
              "          [0.33982825]],\n",
              "\n",
              "         ...,\n",
              "\n",
              "         [[1.0950022 ],\n",
              "          [1.0950022 ],\n",
              "          [1.0950022 ],\n",
              "          ...,\n",
              "          [0.11327608],\n",
              "          [0.11327608],\n",
              "          [0.07551739]],\n",
              "\n",
              "         [[1.0950022 ],\n",
              "          [1.0950022 ],\n",
              "          [1.0950022 ],\n",
              "          ...,\n",
              "          [0.11327608],\n",
              "          [0.11327608],\n",
              "          [0.07551739]],\n",
              "\n",
              "         [[1.0950022 ],\n",
              "          [1.0950022 ],\n",
              "          [1.0950022 ],\n",
              "          ...,\n",
              "          [0.07551739],\n",
              "          [0.07551739],\n",
              "          [0.07551739]]],\n",
              "\n",
              "\n",
              "        [[[1.472589  ],\n",
              "          [1.472589  ],\n",
              "          [1.0572435 ],\n",
              "          ...,\n",
              "          [0.33982825],\n",
              "          [0.30206954],\n",
              "          [0.30206954]],\n",
              "\n",
              "         [[1.359313  ],\n",
              "          [1.472589  ],\n",
              "          [1.1705195 ],\n",
              "          ...,\n",
              "          [0.33982825],\n",
              "          [0.30206954],\n",
              "          [0.30206954]],\n",
              "\n",
              "         [[1.3970717 ],\n",
              "          [1.4348304 ],\n",
              "          [1.472589  ],\n",
              "          ...,\n",
              "          [0.33982825],\n",
              "          [0.33982825],\n",
              "          [0.33982825]],\n",
              "\n",
              "         ...,\n",
              "\n",
              "         [[1.0194848 ],\n",
              "          [1.0194848 ],\n",
              "          [1.0572435 ],\n",
              "          ...,\n",
              "          [0.07551739],\n",
              "          [0.07551739],\n",
              "          [0.03775869]],\n",
              "\n",
              "         [[1.0194848 ],\n",
              "          [1.0194848 ],\n",
              "          [0.98172605],\n",
              "          ...,\n",
              "          [0.07551739],\n",
              "          [0.07551739],\n",
              "          [0.03775869]],\n",
              "\n",
              "         [[1.0194848 ],\n",
              "          [1.0572435 ],\n",
              "          [1.0194848 ],\n",
              "          ...,\n",
              "          [0.03775869],\n",
              "          [0.03775869],\n",
              "          [0.03775869]]]],\n",
              "\n",
              "\n",
              "\n",
              "       [[[[1.3720742 ],\n",
              "          [1.3720742 ],\n",
              "          [1.4112763 ],\n",
              "          ...,\n",
              "          [0.11760636],\n",
              "          [0.07840424],\n",
              "          [0.07840424]],\n",
              "\n",
              "         [[1.5288827 ],\n",
              "          [1.4504786 ],\n",
              "          [1.4112763 ],\n",
              "          ...,\n",
              "          [0.11760636],\n",
              "          [0.07840424],\n",
              "          [0.07840424]],\n",
              "\n",
              "         [[1.4504786 ],\n",
              "          [1.4504786 ],\n",
              "          [1.4504786 ],\n",
              "          ...,\n",
              "          [0.11760636],\n",
              "          [0.11760636],\n",
              "          [0.11760636]],\n",
              "\n",
              "         ...,\n",
              "\n",
              "         [[1.0976593 ],\n",
              "          [1.0584573 ],\n",
              "          [1.0584573 ],\n",
              "          ...,\n",
              "          [9.878935  ],\n",
              "          [9.878935  ],\n",
              "          [9.878935  ]],\n",
              "\n",
              "         [[1.0584573 ],\n",
              "          [1.0584573 ],\n",
              "          [1.0584573 ],\n",
              "          ...,\n",
              "          [9.839732  ],\n",
              "          [9.839732  ],\n",
              "          [9.839732  ]],\n",
              "\n",
              "         [[1.0584573 ],\n",
              "          [1.0584573 ],\n",
              "          [1.0584573 ],\n",
              "          ...,\n",
              "          [9.839732  ],\n",
              "          [9.839732  ],\n",
              "          [9.839732  ]]],\n",
              "\n",
              "\n",
              "        [[[1.4504786 ],\n",
              "          [1.5680848 ],\n",
              "          [1.5288827 ],\n",
              "          ...,\n",
              "          [0.07840424],\n",
              "          [0.07840424],\n",
              "          [0.07840424]],\n",
              "\n",
              "         [[1.4504786 ],\n",
              "          [1.5680848 ],\n",
              "          [1.5680848 ],\n",
              "          ...,\n",
              "          [0.07840424],\n",
              "          [0.07840424],\n",
              "          [0.07840424]],\n",
              "\n",
              "         [[1.4504786 ],\n",
              "          [1.4504786 ],\n",
              "          [1.4112763 ],\n",
              "          ...,\n",
              "          [0.07840424],\n",
              "          [0.07840424],\n",
              "          [0.07840424]],\n",
              "\n",
              "         ...,\n",
              "\n",
              "         [[1.0584573 ],\n",
              "          [1.0584573 ],\n",
              "          [1.0192552 ],\n",
              "          ...,\n",
              "          [9.80053   ],\n",
              "          [9.80053   ],\n",
              "          [9.80053   ]],\n",
              "\n",
              "         [[1.0584573 ],\n",
              "          [1.0584573 ],\n",
              "          [1.0584573 ],\n",
              "          ...,\n",
              "          [9.80053   ],\n",
              "          [9.80053   ],\n",
              "          [9.80053   ]],\n",
              "\n",
              "         [[1.0584573 ],\n",
              "          [1.0584573 ],\n",
              "          [1.0584573 ],\n",
              "          ...,\n",
              "          [9.80053   ],\n",
              "          [9.80053   ],\n",
              "          [9.80053   ]]],\n",
              "\n",
              "\n",
              "        [[[1.5288827 ],\n",
              "          [1.5288827 ],\n",
              "          [1.4504786 ],\n",
              "          ...,\n",
              "          [0.07840424],\n",
              "          [0.07840424],\n",
              "          [0.07840424]],\n",
              "\n",
              "         [[1.5288827 ],\n",
              "          [1.5288827 ],\n",
              "          [1.5288827 ],\n",
              "          ...,\n",
              "          [0.07840424],\n",
              "          [0.07840424],\n",
              "          [0.07840424]],\n",
              "\n",
              "         [[1.4504786 ],\n",
              "          [1.4504786 ],\n",
              "          [1.4112763 ],\n",
              "          ...,\n",
              "          [0.07840424],\n",
              "          [0.07840424],\n",
              "          [0.07840424]],\n",
              "\n",
              "         ...,\n",
              "\n",
              "         [[1.0584573 ],\n",
              "          [1.0584573 ],\n",
              "          [1.0192552 ],\n",
              "          ...,\n",
              "          [9.80053   ],\n",
              "          [9.80053   ],\n",
              "          [9.80053   ]],\n",
              "\n",
              "         [[1.0584573 ],\n",
              "          [1.0584573 ],\n",
              "          [1.0584573 ],\n",
              "          ...,\n",
              "          [9.80053   ],\n",
              "          [9.80053   ],\n",
              "          [9.80053   ]],\n",
              "\n",
              "         [[1.0584573 ],\n",
              "          [1.0584573 ],\n",
              "          [1.0584573 ],\n",
              "          ...,\n",
              "          [9.839732  ],\n",
              "          [9.839732  ],\n",
              "          [9.80053   ]]],\n",
              "\n",
              "\n",
              "        ...,\n",
              "\n",
              "\n",
              "        [[[1.4896806 ],\n",
              "          [1.5680848 ],\n",
              "          [1.5288827 ],\n",
              "          ...,\n",
              "          [0.11760636],\n",
              "          [0.11760636],\n",
              "          [0.11760636]],\n",
              "\n",
              "         [[1.4504786 ],\n",
              "          [1.5680848 ],\n",
              "          [1.607287  ],\n",
              "          ...,\n",
              "          [0.11760636],\n",
              "          [0.11760636],\n",
              "          [0.11760636]],\n",
              "\n",
              "         [[1.4504786 ],\n",
              "          [1.4504786 ],\n",
              "          [1.4504786 ],\n",
              "          ...,\n",
              "          [0.11760636],\n",
              "          [0.11760636],\n",
              "          [0.11760636]],\n",
              "\n",
              "         ...,\n",
              "\n",
              "         [[1.0192552 ],\n",
              "          [1.0192552 ],\n",
              "          [1.0192552 ],\n",
              "          ...,\n",
              "          [9.878935  ],\n",
              "          [9.878935  ],\n",
              "          [9.878935  ]],\n",
              "\n",
              "         [[1.0192552 ],\n",
              "          [1.0192552 ],\n",
              "          [1.0192552 ],\n",
              "          ...,\n",
              "          [9.878935  ],\n",
              "          [9.878935  ],\n",
              "          [9.878935  ]],\n",
              "\n",
              "         [[1.0192552 ],\n",
              "          [1.0192552 ],\n",
              "          [0.98005307],\n",
              "          ...,\n",
              "          [9.878935  ],\n",
              "          [9.878935  ],\n",
              "          [9.878935  ]]],\n",
              "\n",
              "\n",
              "        [[[1.4896806 ],\n",
              "          [1.5680848 ],\n",
              "          [1.5680848 ],\n",
              "          ...,\n",
              "          [0.11760636],\n",
              "          [0.11760636],\n",
              "          [0.11760636]],\n",
              "\n",
              "         [[1.4504786 ],\n",
              "          [1.5680848 ],\n",
              "          [1.6464891 ],\n",
              "          ...,\n",
              "          [0.11760636],\n",
              "          [0.11760636],\n",
              "          [0.07840424]],\n",
              "\n",
              "         [[1.4504786 ],\n",
              "          [1.4504786 ],\n",
              "          [1.4504786 ],\n",
              "          ...,\n",
              "          [0.11760636],\n",
              "          [0.11760636],\n",
              "          [0.11760636]],\n",
              "\n",
              "         ...,\n",
              "\n",
              "         [[1.0584573 ],\n",
              "          [1.0584573 ],\n",
              "          [1.0584573 ],\n",
              "          ...,\n",
              "          [9.918137  ],\n",
              "          [9.918137  ],\n",
              "          [9.918137  ]],\n",
              "\n",
              "         [[1.0584573 ],\n",
              "          [1.0584573 ],\n",
              "          [1.0584573 ],\n",
              "          ...,\n",
              "          [9.878935  ],\n",
              "          [9.878935  ],\n",
              "          [9.878935  ]],\n",
              "\n",
              "         [[1.0584573 ],\n",
              "          [1.0584573 ],\n",
              "          [1.0192552 ],\n",
              "          ...,\n",
              "          [9.878935  ],\n",
              "          [9.878935  ],\n",
              "          [9.878935  ]]],\n",
              "\n",
              "\n",
              "        [[[1.5288827 ],\n",
              "          [1.5288827 ],\n",
              "          [1.4504786 ],\n",
              "          ...,\n",
              "          [0.07840424],\n",
              "          [0.11760636],\n",
              "          [0.11760636]],\n",
              "\n",
              "         [[1.4504786 ],\n",
              "          [1.4504786 ],\n",
              "          [1.4504786 ],\n",
              "          ...,\n",
              "          [0.07840424],\n",
              "          [0.11760636],\n",
              "          [0.07840424]],\n",
              "\n",
              "         [[1.4504786 ],\n",
              "          [1.4504786 ],\n",
              "          [1.4112763 ],\n",
              "          ...,\n",
              "          [0.15680848],\n",
              "          [0.15680848],\n",
              "          [0.15680848]],\n",
              "\n",
              "         ...,\n",
              "\n",
              "         [[1.0584573 ],\n",
              "          [1.0584573 ],\n",
              "          [1.0584573 ],\n",
              "          ...,\n",
              "          [9.839732  ],\n",
              "          [9.839732  ],\n",
              "          [9.839732  ]],\n",
              "\n",
              "         [[1.0584573 ],\n",
              "          [1.0584573 ],\n",
              "          [1.0584573 ],\n",
              "          ...,\n",
              "          [9.839732  ],\n",
              "          [9.839732  ],\n",
              "          [9.839732  ]],\n",
              "\n",
              "         [[1.0584573 ],\n",
              "          [1.0584573 ],\n",
              "          [1.0192552 ],\n",
              "          ...,\n",
              "          [9.80053   ],\n",
              "          [9.80053   ],\n",
              "          [9.80053   ]]]]], dtype=float32)"
            ]
          },
          "metadata": {},
          "execution_count": 181
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "imageio.mimsave('animation.gif',val[0][0],fps=10) \n",
        "#Cool function alert! This converts the first video instance out of the 2 in the batch to a gif of 10 fps."
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "S1nMO7stWPV_",
        "outputId": "af1818b6-3bf2-4a7f-ee80-0fbd772dff1e"
      },
      "execution_count": 182,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n",
            "WARNING:imageio:Lossy conversion from float32 to uint8. Range [0.0, 9.628466606140137]. Convert image to uint8 prior to saving to suppress this warning.\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "So basically the gif created is what the model trains on along with the alignemnts to perform lip reading."
      ],
      "metadata": {
        "id": "BtaQcCwwWSyE"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# 0/1: videos/alignments, 0/1: first or second video, 0-74: frames of video 1.\n",
        "plt.imshow(val[0][0][0])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 244
        },
        "id": "0QaOMr47WUBt",
        "outputId": "3b6369d0-fded-4fed-b1b3-e75c1a0f86b0"
      },
      "execution_count": 183,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7ff5f8304070>"
            ]
          },
          "metadata": {},
          "execution_count": 183
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "tf.strings.reduce_join([num_to_char(word) for word in val[1][0]])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "dmn2vBSUWfKZ",
        "outputId": "cda352ce-d7e1-4272-b6af-e19a8fc4f23d"
      },
      "execution_count": 184,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tf.Tensor: shape=(), dtype=string, numpy=b'place white by x nine soon'>"
            ]
          },
          "metadata": {},
          "execution_count": 184
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from tensorflow.keras.models import Sequential\n",
        "from tensorflow.keras.layers import Dense,Conv3D,LSTM, Dropout,Bidirectional, Activation, SpatialDropout3D, BatchNormalization, MaxPool3D,Reshape, Flatten,TimeDistributed\n",
        "from tensorflow.keras.optimizers import Adam\n",
        "from tensorflow.keras.callbacks import ModelCheckpoint, LearningRateScheduler"
      ],
      "metadata": {
        "id": "8btFcGs_Whrd"
      },
      "execution_count": 185,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "data.as_numpy_iterator().next()[0][0].shape"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "byFf58BwWntF",
        "outputId": "1aaf462e-6fa9-4498-af8f-07f690abc3da"
      },
      "execution_count": 186,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(75, 46, 140, 1)"
            ]
          },
          "metadata": {},
          "execution_count": 186
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "char_to_num.vocabulary_size()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "yUm2GlhhWpru",
        "outputId": "c5059a00-0d91-4499-cb0e-084b197392ac"
      },
      "execution_count": 187,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "40"
            ]
          },
          "metadata": {},
          "execution_count": 187
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "model = Sequential()\n",
        "model.add(Conv3D(128, 3, input_shape=(75,46,140,1), padding='same'))\n",
        "model.add(Activation('relu'))\n",
        "model.add(MaxPool3D((1,2,2)))\n",
        "\n",
        "model.add(Conv3D(256, 3, padding='same'))\n",
        "model.add(Activation('relu'))\n",
        "model.add(MaxPool3D((1,2,2)))\n",
        "\n",
        "model.add(Conv3D(75, 3, padding='same'))\n",
        "model.add(Activation('relu'))\n",
        "model.add(MaxPool3D((1,2,2)))\n",
        "\n",
        "model.add(TimeDistributed(Flatten()))\n",
        "\n",
        "model.add(Bidirectional(LSTM(128, kernel_initializer='Orthogonal', return_sequences=True)))\n",
        "model.add(Dropout(.5))\n",
        "\n",
        "model.add(Bidirectional(LSTM(128, kernel_initializer='Orthogonal', return_sequences=True)))\n",
        "model.add(Dropout(.5))\n",
        "\n",
        "model.add(Dense(char_to_num.vocabulary_size()+1, kernel_initializer='he_normal', activation='softmax'))"
      ],
      "metadata": {
        "id": "SKnhup9bWsja"
      },
      "execution_count": 188,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "model.summary()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "YFQM8o7WWwkv",
        "outputId": "ec8d37ab-dab1-496e-ef04-1fcbe9aba9e1"
      },
      "execution_count": 189,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_3\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " conv3d_9 (Conv3D)           (None, 75, 46, 140, 128)  3584      \n",
            "                                                                 \n",
            " activation_9 (Activation)   (None, 75, 46, 140, 128)  0         \n",
            "                                                                 \n",
            " max_pooling3d_9 (MaxPooling  (None, 75, 23, 70, 128)  0         \n",
            " 3D)                                                             \n",
            "                                                                 \n",
            " conv3d_10 (Conv3D)          (None, 75, 23, 70, 256)   884992    \n",
            "                                                                 \n",
            " activation_10 (Activation)  (None, 75, 23, 70, 256)   0         \n",
            "                                                                 \n",
            " max_pooling3d_10 (MaxPoolin  (None, 75, 11, 35, 256)  0         \n",
            " g3D)                                                            \n",
            "                                                                 \n",
            " conv3d_11 (Conv3D)          (None, 75, 11, 35, 75)    518475    \n",
            "                                                                 \n",
            " activation_11 (Activation)  (None, 75, 11, 35, 75)    0         \n",
            "                                                                 \n",
            " max_pooling3d_11 (MaxPoolin  (None, 75, 5, 17, 75)    0         \n",
            " g3D)                                                            \n",
            "                                                                 \n",
            " time_distributed_3 (TimeDis  (None, 75, 6375)         0         \n",
            " tributed)                                                       \n",
            "                                                                 \n",
            " bidirectional_6 (Bidirectio  (None, 75, 256)          6660096   \n",
            " nal)                                                            \n",
            "                                                                 \n",
            " dropout_6 (Dropout)         (None, 75, 256)           0         \n",
            "                                                                 \n",
            " bidirectional_7 (Bidirectio  (None, 75, 256)          394240    \n",
            " nal)                                                            \n",
            "                                                                 \n",
            " dropout_7 (Dropout)         (None, 75, 256)           0         \n",
            "                                                                 \n",
            " dense_3 (Dense)             (None, 75, 41)            10537     \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 8,471,924\n",
            "Trainable params: 8,471,924\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "yhat = model.predict(val[0])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "B_OHK7P_W0jv",
        "outputId": "2f9516e6-a981-4321-e16f-efe4b065923c"
      },
      "execution_count": 190,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "1/1 [==============================] - 2s 2s/step\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "yhat[0].shape"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "-mlmpo_nW1IX",
        "outputId": "8f3db2a1-0955-4515-e6ec-8990b1fb573e"
      },
      "execution_count": 191,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(75, 41)"
            ]
          },
          "metadata": {},
          "execution_count": 191
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "tf.strings.reduce_join([num_to_char(tf.argmax(x)) for x in yhat[1]])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "yDK7_fm5W4QE",
        "outputId": "22a4d3ed-c97e-4d9b-c3c8-893c336f0860"
      },
      "execution_count": 192,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tf.Tensor: shape=(), dtype=string, numpy=b'999944444444444444444443333344444444444444444444444444444444444444444444444'>"
            ]
          },
          "metadata": {},
          "execution_count": 192
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "model.input_shape"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "fkp0FmQ1W58_",
        "outputId": "b0d76f91-d2b2-4793-b9f4-a560d7e23221"
      },
      "execution_count": 193,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(None, 75, 46, 140, 1)"
            ]
          },
          "metadata": {},
          "execution_count": 193
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "model.output_shape"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "f-CtqFjZW871",
        "outputId": "d9dab830-80d1-4271-ea09-aa5692e4a42f"
      },
      "execution_count": 194,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(None, 75, 41)"
            ]
          },
          "metadata": {},
          "execution_count": 194
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Setting Up Training Options and Train"
      ],
      "metadata": {
        "id": "2DC1bxSfW_q_"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "def scheduler(epoch,lr):\n",
        "  if epoch<30:\n",
        "    return lr\n",
        "  else:\n",
        "    return lr*tf.math.exp(-0.1)"
      ],
      "metadata": {
        "id": "i7CecBhNW-Wg"
      },
      "execution_count": 195,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#CTC LOSS - this block of code has been picked up from Automatic Speech Recognition using CTC paper.\n",
        "def CTCLoss(y_true,y_pred):\n",
        "  batch_len = tf.cast(tf.shape(y_true)[0], dtype='int64')\n",
        "  input_length = tf.cast(tf.shape(y_pred)[1], dtype='int64')\n",
        "  label_length = tf.cast(tf.shape(y_true)[1], dtype='int64')\n",
        "\n",
        "  input_length=input_length* tf.ones(shape=(batch_len,1),dtype='int64')\n",
        "  label_length = label_length* tf.ones(shape=(batch_len,1), dtype='int64')\n",
        "\n",
        "  loss = tf.keras.backend.ctc_batch_cost(y_true,y_pred,input_length, label_length)\n",
        "  return loss"
      ],
      "metadata": {
        "id": "n1yXfX2EXIbx"
      },
      "execution_count": 196,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "class ProduceExample(tf.keras.callbacks.Callback): \n",
        "    def __init__(self, dataset) -> None: \n",
        "        self.dataset = dataset.as_numpy_iterator()\n",
        "    \n",
        "    def on_epoch_end(self, epoch, logs=None) -> None:\n",
        "        data = self.dataset.next()\n",
        "        yhat = self.model.predict(data[0])\n",
        "        decoded = tf.keras.backend.ctc_decode(yhat, [75,75], greedy=False)[0][0].numpy()\n",
        "        for x in range(len(yhat)):           \n",
        "            print('Original:', tf.strings.reduce_join(num_to_char(data[1][x])).numpy().decode('utf-8'))\n",
        "            print('Prediction:', tf.strings.reduce_join(num_to_char(decoded[x])).numpy().decode('utf-8'))\n",
        "            print('~'*100)"
      ],
      "metadata": {
        "id": "AdQhi8aeXJ18"
      },
      "execution_count": 197,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "model.compile(optimizer=tf.keras.optimizers.legacy.Adam(learning_rate=0.0001),loss =CTCLoss)"
      ],
      "metadata": {
        "id": "PHxGTvQ8XLXI"
      },
      "execution_count": 198,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "checkpoint_callback= ModelCheckpoint(os.path.join('models','checkpoint'),monitor='loss', save_weights_only=True)\n",
        "                                     "
      ],
      "metadata": {
        "id": "EjwDP6RtXO4l"
      },
      "execution_count": 199,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "schedule_callback = LearningRateScheduler(scheduler)"
      ],
      "metadata": {
        "id": "FwF2kWckXQJa"
      },
      "execution_count": 200,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "example_callback = ProduceExample(data)"
      ],
      "metadata": {
        "id": "P6IETmozXRcE"
      },
      "execution_count": 201,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "model.fit(train, validation_data=test, epochs=100, callbacks=[checkpoint_callback, schedule_callback, example_callback], )"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 592
        },
        "id": "Z2biqHzEXSq3",
        "outputId": "17a83d3d-f2a1-4a66-b05c-3acd45b3ec5c"
      },
      "execution_count": 202,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/100\n",
            "1/1 [==============================] - 2s 2s/step\n",
            "Original: lay green at z seven soon\n",
            "Prediction: le e e e e o\n",
            "~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\n",
            "Original: lay blue at q six please\n",
            "Prediction: le e e e e e o\n",
            "~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\n",
            "450/450 [==============================] - 709s 2s/step - loss: 85.4129 - val_loss: 71.2436 - lr: 1.0000e-04\n",
            "Epoch 2/100\n",
            "171/450 [==========>...................] - ETA: 4:23 - loss: 73.5683"
          ]
        },
        {
          "output_type": "error",
          "ename": "KeyboardInterrupt",
          "evalue": "ignored",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-202-f338ee738322>\u001b[0m in \u001b[0;36m<cell line: 1>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mmodel\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtrain\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mvalidation_data\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mtest\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mepochs\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m100\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcallbacks\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mcheckpoint_callback\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mschedule_callback\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mexample_callback\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
            "\u001b[0;32m/usr/local/lib/python3.10/dist-packages/keras/utils/traceback_utils.py\u001b[0m in \u001b[0;36merror_handler\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m     63\u001b[0m         \u001b[0mfiltered_tb\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     64\u001b[0m         \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 65\u001b[0;31m             \u001b[0;32mreturn\u001b[0m \u001b[0mfn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     66\u001b[0m         \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     67\u001b[0m             \u001b[0mfiltered_tb\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0m_process_traceback_frames\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m__traceback__\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.10/dist-packages/keras/engine/training.py\u001b[0m in \u001b[0;36mfit\u001b[0;34m(self, x, y, batch_size, epochs, verbose, callbacks, validation_split, validation_data, shuffle, class_weight, sample_weight, initial_epoch, steps_per_epoch, validation_steps, validation_batch_size, validation_freq, max_queue_size, workers, use_multiprocessing)\u001b[0m\n\u001b[1;32m   1689\u001b[0m                             \u001b[0mlogs\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtmp_logs\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1690\u001b[0m                             \u001b[0mend_step\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mstep\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0mdata_handler\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstep_increment\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1691\u001b[0;31m                             \u001b[0mcallbacks\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mon_train_batch_end\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mend_step\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlogs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1692\u001b[0m                             \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstop_training\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1693\u001b[0m                                 \u001b[0;32mbreak\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.10/dist-packages/keras/callbacks.py\u001b[0m in \u001b[0;36mon_train_batch_end\u001b[0;34m(self, batch, logs)\u001b[0m\n\u001b[1;32m    473\u001b[0m         \"\"\"\n\u001b[1;32m    474\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_should_call_train_batch_hooks\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 475\u001b[0;31m             \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_call_batch_hook\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mModeKeys\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mTRAIN\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m\"end\"\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mbatch\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlogs\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mlogs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    476\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    477\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mon_test_batch_begin\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mbatch\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlogs\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mNone\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.10/dist-packages/keras/callbacks.py\u001b[0m in \u001b[0;36m_call_batch_hook\u001b[0;34m(self, mode, hook, batch, logs)\u001b[0m\n\u001b[1;32m    320\u001b[0m             \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_call_batch_begin_hook\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mmode\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mbatch\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlogs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    321\u001b[0m         \u001b[0;32melif\u001b[0m \u001b[0mhook\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;34m\"end\"\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 322\u001b[0;31m             \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_call_batch_end_hook\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mmode\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mbatch\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlogs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    323\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    324\u001b[0m             raise ValueError(\n",
            "\u001b[0;32m/usr/local/lib/python3.10/dist-packages/keras/callbacks.py\u001b[0m in \u001b[0;36m_call_batch_end_hook\u001b[0;34m(self, mode, batch, logs)\u001b[0m\n\u001b[1;32m    343\u001b[0m             \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_batch_times\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mbatch_time\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    344\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 345\u001b[0;31m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_call_batch_hook_helper\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mhook_name\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mbatch\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlogs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    346\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    347\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mlen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_batch_times\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m>=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_num_batches_for_timing_check\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.10/dist-packages/keras/callbacks.py\u001b[0m in \u001b[0;36m_call_batch_hook_helper\u001b[0;34m(self, hook_name, batch, logs)\u001b[0m\n\u001b[1;32m    391\u001b[0m         \u001b[0;32mfor\u001b[0m \u001b[0mcallback\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcallbacks\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    392\u001b[0m             \u001b[0mhook\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mgetattr\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mcallback\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mhook_name\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 393\u001b[0;31m             \u001b[0mhook\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mbatch\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlogs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    394\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    395\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_check_timing\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.10/dist-packages/keras/callbacks.py\u001b[0m in \u001b[0;36mon_train_batch_end\u001b[0;34m(self, batch, logs)\u001b[0m\n\u001b[1;32m   1091\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1092\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mon_train_batch_end\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mbatch\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlogs\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mNone\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1093\u001b[0;31m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_batch_update_progbar\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mbatch\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlogs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1094\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1095\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mon_test_batch_end\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mbatch\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlogs\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mNone\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.10/dist-packages/keras/callbacks.py\u001b[0m in \u001b[0;36m_batch_update_progbar\u001b[0;34m(self, batch, logs)\u001b[0m\n\u001b[1;32m   1167\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mverbose\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;36m1\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1168\u001b[0m             \u001b[0;31m# Only block async when verbose = 1.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1169\u001b[0;31m             \u001b[0mlogs\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtf_utils\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msync_to_numpy_or_python_type\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlogs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1170\u001b[0m             \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mprogbar\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mupdate\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mseen\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlist\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlogs\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mitems\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfinalize\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mFalse\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1171\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.10/dist-packages/keras/utils/tf_utils.py\u001b[0m in \u001b[0;36msync_to_numpy_or_python_type\u001b[0;34m(tensors)\u001b[0m\n\u001b[1;32m    678\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mt\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mitem\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mif\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mndim\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mt\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;36m0\u001b[0m \u001b[0;32melse\u001b[0m \u001b[0mt\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    679\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 680\u001b[0;31m     \u001b[0;32mreturn\u001b[0m \u001b[0mtf\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mnest\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmap_structure\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0m_to_single_numpy_or_python_type\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtensors\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    681\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    682\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.10/dist-packages/tensorflow/python/util/nest.py\u001b[0m in \u001b[0;36mmap_structure\u001b[0;34m(func, *structure, **kwargs)\u001b[0m\n\u001b[1;32m    915\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    916\u001b[0m   return pack_sequence_as(\n\u001b[0;32m--> 917\u001b[0;31m       \u001b[0mstructure\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0mfunc\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mx\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mentries\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    918\u001b[0m       expand_composites=expand_composites)\n\u001b[1;32m    919\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.10/dist-packages/tensorflow/python/util/nest.py\u001b[0m in \u001b[0;36m<listcomp>\u001b[0;34m(.0)\u001b[0m\n\u001b[1;32m    915\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    916\u001b[0m   return pack_sequence_as(\n\u001b[0;32m--> 917\u001b[0;31m       \u001b[0mstructure\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0mfunc\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mx\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mentries\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    918\u001b[0m       expand_composites=expand_composites)\n\u001b[1;32m    919\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.10/dist-packages/keras/utils/tf_utils.py\u001b[0m in \u001b[0;36m_to_single_numpy_or_python_type\u001b[0;34m(t)\u001b[0m\n\u001b[1;32m    671\u001b[0m         \u001b[0;31m# Don't turn ragged or sparse tensors to NumPy.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    672\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0misinstance\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mt\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtf\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mTensor\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 673\u001b[0;31m             \u001b[0mt\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mt\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mnumpy\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    674\u001b[0m         \u001b[0;31m# Strings, ragged and sparse tensors don't have .item(). Return them\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    675\u001b[0m         \u001b[0;31m# as-is.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.10/dist-packages/tensorflow/python/framework/ops.py\u001b[0m in \u001b[0;36mnumpy\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m   1158\u001b[0m     \"\"\"\n\u001b[1;32m   1159\u001b[0m     \u001b[0;31m# TODO(slebedev): Consider avoiding a copy for non-CPU or remote tensors.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1160\u001b[0;31m     \u001b[0mmaybe_arr\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_numpy\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m  \u001b[0;31m# pylint: disable=protected-access\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1161\u001b[0m     \u001b[0;32mreturn\u001b[0m \u001b[0mmaybe_arr\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcopy\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mif\u001b[0m \u001b[0misinstance\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mmaybe_arr\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mndarray\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32melse\u001b[0m \u001b[0mmaybe_arr\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1162\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.10/dist-packages/tensorflow/python/framework/ops.py\u001b[0m in \u001b[0;36m_numpy\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m   1124\u001b[0m   \u001b[0;32mdef\u001b[0m \u001b[0m_numpy\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1125\u001b[0m     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1126\u001b[0;31m       \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_numpy_internal\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1127\u001b[0m     \u001b[0;32mexcept\u001b[0m \u001b[0mcore\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_NotOkStatusException\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m  \u001b[0;31m# pylint: disable=protected-access\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1128\u001b[0m       \u001b[0;32mraise\u001b[0m \u001b[0mcore\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_status_to_exception\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0me\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0;32mNone\u001b[0m  \u001b[0;31m# pylint: disable=protected-access\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "as we reach 50+ epochs we see the model working out some comparable predictions."
      ],
      "metadata": {
        "id": "9XKJctnhzP13"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Making Predictions"
      ],
      "metadata": {
        "id": "pl1-3Aj6gXNY"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Training for a 100 epochs takes alot of time. So i downloaded the model checkpoints available after 96 epochs of the above model training."
      ],
      "metadata": {
        "id": "nvYyTsCmqVVR"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "url = 'https://drive.google.com/uc?id=1vWscXs4Vt0a_1IH1-ct2TCgXAZT-N3_Y'\n",
        "output = 'checkpoints.zip'\n",
        "gdown.download(url, output, quiet=False)\n",
        "gdown.extractall('checkpoints.zip', 'models')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "TkX-15huX10X",
        "outputId": "93e7c16d-179a-469b-cd29-736ab6f72a84"
      },
      "execution_count": 203,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "Downloading...\n",
            "From: https://drive.google.com/uc?id=1vWscXs4Vt0a_1IH1-ct2TCgXAZT-N3_Y\n",
            "To: /content/checkpoints.zip\n",
            "100%|██████████| 94.5M/94.5M [00:00<00:00, 125MB/s]\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['models/checkpoint.index',\n",
              " 'models/__MACOSX/._checkpoint.index',\n",
              " 'models/checkpoint.data-00000-of-00001',\n",
              " 'models/__MACOSX/._checkpoint.data-00000-of-00001',\n",
              " 'models/checkpoint',\n",
              " 'models/__MACOSX/._checkpoint']"
            ]
          },
          "metadata": {},
          "execution_count": 203
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "model.load_weights('models/checkpoint')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "07UTirFWv0NP",
        "outputId": "cd4a823e-16ab-498f-e8a1-6f660ed07a99"
      },
      "execution_count": 204,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tensorflow.python.checkpoint.checkpoint.CheckpointLoadStatus at 0x7ff5deff7b20>"
            ]
          },
          "metadata": {},
          "execution_count": 204
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "sample = test.as_numpy_iterator().next()"
      ],
      "metadata": {
        "id": "CM0W0MSLwItr"
      },
      "execution_count": 205,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "yhat = model.predict(sample[0])"
      ],
      "metadata": {
        "id": "v1Iznxlksrrs",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "78d615db-9be9-4ca0-f044-4b590cc2013b"
      },
      "execution_count": 206,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "1/1 [==============================] - 0s 263ms/step\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "print('~'*100, 'REAL TEXT')\n",
        "[tf.strings.reduce_join([num_to_char(word) for word in sentence]) for sentence in sample[1]]"
      ],
      "metadata": {
        "id": "6zfUpl_K3hXm",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "64ad3431-1300-4d2d-8b9e-abff591bc046"
      },
      "execution_count": 207,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ REAL TEXT\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[<tf.Tensor: shape=(), dtype=string, numpy=b'lay blue in d three soon'>,\n",
              " <tf.Tensor: shape=(), dtype=string, numpy=b'set red at o zero please'>]"
            ]
          },
          "metadata": {},
          "execution_count": 207
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "decoded = tf.keras.backend.ctc_decode(yhat, input_length=[75,75], greedy=True)[0][0].numpy()\n",
        "print('~'*100, 'PREDICTIONS')\n",
        "[tf.strings.reduce_join([num_to_char(word) for word in sentence]) for sentence in decoded]\n"
      ],
      "metadata": {
        "id": "TYYUfjF44aVo",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "6a35e62f-d361-410b-cbed-6b7ab7e3d5ea"
      },
      "execution_count": 208,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ PREDICTIONS\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[<tf.Tensor: shape=(), dtype=string, numpy=b'lay blue in d three soon'>,\n",
              " <tf.Tensor: shape=(), dtype=string, numpy=b'set red at z zero please'>]"
            ]
          },
          "metadata": {},
          "execution_count": 208
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Our predictions seem to be near perfect and the model seems to be performing nicely."
      ],
      "metadata": {
        "id": "WpGwQ1cdu5pD"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "## TEST ON A VIDEO"
      ],
      "metadata": {
        "id": "ZAMkwxPd4jbo"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "sample= load_data(tf.convert_to_tensor('/content/data/s1/briz8p.mpg'))"
      ],
      "metadata": {
        "id": "_5fgsDyT4gm6"
      },
      "execution_count": 214,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "print('~'*100, 'REAL TEXT')\n",
        "[tf.strings.reduce_join([num_to_char(word) for word in sentence]) for sentence in [sample[1]]]"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "XC0RtvTB0q9j",
        "outputId": "8988a88d-7c07-4d19-b2f6-3ea9aef2a049"
      },
      "execution_count": 215,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ REAL TEXT\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[<tf.Tensor: shape=(), dtype=string, numpy=b'bin red in z eight please'>]"
            ]
          },
          "metadata": {},
          "execution_count": 215
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "yhat = model.predict(tf.expand_dims(sample[0], axis=0))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "xn3gRxxE0vej",
        "outputId": "a98af95e-b80a-4c68-8db9-801af289c682"
      },
      "execution_count": 216,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "1/1 [==============================] - 0s 166ms/step\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "decoded = tf.keras.backend.ctc_decode(yhat, input_length=[75], greedy=True)[0][0].numpy()"
      ],
      "metadata": {
        "id": "RfS3dc9801va"
      },
      "execution_count": 217,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "print('~'*100, 'PREDICTIONS')\n",
        "[tf.strings.reduce_join([num_to_char(word) for word in sentence]) for sentence in decoded]"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "WIRF3BTP05LP",
        "outputId": "a844f218-a34e-473d-e8e1-202805e99b7e"
      },
      "execution_count": 218,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ PREDICTIONS\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[<tf.Tensor: shape=(), dtype=string, numpy=b'bin red in z eight please'>]"
            ]
          },
          "metadata": {},
          "execution_count": 218
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [],
      "metadata": {
        "id": "ypDe9UXc1AKv"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "nice predictions. good results."
      ],
      "metadata": {
        "id": "lYEdkacn1BXw"
      }
    }
  ]
}