sentiment.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2023-10-15 03:05:47.954148: I tensorflow/tsl/cuda/cudart_stub.cc:28] Could not find cuda drivers on your machine, GPU will not be used.\n",
      "2023-10-15 03:05:47.993905: I tensorflow/tsl/cuda/cudart_stub.cc:28] Could not find cuda drivers on your machine, GPU will not be used.\n",
      "2023-10-15 03:05:47.994394: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.\n",
      "To enable the following instructions: AVX2 FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.\n",
      "2023-10-15 03:05:48.726586: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "import scipy.io.wavfile as wav\n",
    "from speechpy.feature import mfcc\n",
    "import tensorflow as tf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import sys\n",
    "from typing import Tuple"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "mean_signal_length = 32000\n",
    "# mean_signal_length = 88500"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_feature_vector_from_mfcc(file_path: str, flatten: bool, mfcc_len: int = 39) -> np.ndarray:\n",
    "    \"\"\"\n",
    "    Make feature vector from MFCC for the given wav file.\n",
    "\n",
    "    Args:\n",
    "        file_path (str): path to the .wav file that needs to be read.\n",
    "        flatten (bool) : Boolean indicating whether to flatten mfcc obtained.\n",
    "        mfcc_len (int): Number of cepestral co efficients to be consider.\n",
    "\n",
    "    Returns:\n",
    "        numpy.ndarray: feature vector of the wav file made from mfcc.\n",
    "    \"\"\"\n",
    "    fs, signal = wav.read(file_path)\n",
    "    s_len = len(signal)\n",
    "    print(s_len)\n",
    "    # pad the signals to have same size if lesser than required\n",
    "    # else slice them\n",
    "    if s_len < mean_signal_length:\n",
    "        pad_len = mean_signal_length - s_len\n",
    "        pad_rem = pad_len % 2\n",
    "        pad_len //= 2\n",
    "        signal = np.pad(signal, (pad_len, pad_len + pad_rem),\n",
    "                        'constant', constant_values=0)\n",
    "    else:\n",
    "        pad_len = s_len - mean_signal_length\n",
    "        pad_len //= 2\n",
    "        signal = signal[pad_len:pad_len + mean_signal_length]\n",
    "    print(len(signal))\n",
    "    mel_coefficients = mfcc(signal, fs, num_cepstral=mfcc_len)\n",
    "    print(len(signal))\n",
    "    if flatten:\n",
    "        # Flatten the data\n",
    "        mel_coefficients = np.ravel(mel_coefficients)\n",
    "    return mel_coefficients"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "filename = './test.wav'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "132096\n",
      "32000\n",
      "32000\n"
     ]
    }
   ],
   "source": [
    "features = get_feature_vector_from_mfcc(filename, flatten=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(70, 39)"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "features.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_data(data_path: str, flatten: bool = True, mfcc_len: int = 39,\n",
    "             class_labels: Tuple = (\"Neutral\", \"Angry\", \"Happy\", \"Sad\")) -> \\\n",
    "        Tuple[np.ndarray, np.ndarray]:\n",
    "    \"\"\"Extract data for training and testing.\n",
    "\n",
    "    1. Iterate through all the folders.\n",
    "    2. Read the audio files in each folder.\n",
    "    3. Extract Mel frequency cepestral coefficients for each file.\n",
    "    4. Generate feature vector for the audio files as required.\n",
    "\n",
    "    Args:\n",
    "        data_path (str): path to the data set folder\n",
    "        flatten (bool): Boolean specifying whether to flatten the data or not.\n",
    "        mfcc_len (int): Number of mfcc features to take for each frame.\n",
    "        class_labels (tuple): class labels that we care about.\n",
    "\n",
    "    Returns:\n",
    "        Tuple[numpy.ndarray, numpy.ndarray]: Two numpy arrays, one with mfcc and\n",
    "        other with labels.\n",
    "\n",
    "\n",
    "    \"\"\"\n",
    "    data = []\n",
    "    labels = []\n",
    "    names = []\n",
    "    cur_dir = os.getcwd()\n",
    "    sys.stderr.write('curdir: %s\\n' % cur_dir)\n",
    "    os.chdir(data_path)\n",
    "    for i, directory in enumerate(class_labels):\n",
    "        sys.stderr.write(\"started reading folder %s\\n\" % directory)\n",
    "        os.chdir(directory)\n",
    "        for filename in os.listdir('.'):\n",
    "            filepath = os.getcwd() + '/' + filename\n",
    "            feature_vector = get_feature_vector_from_mfcc(file_path=filepath,\n",
    "                                                          mfcc_len=mfcc_len,\n",
    "                                                          flatten=flatten)\n",
    "            data.append(feature_vector)\n",
    "            labels.append(i)\n",
    "            names.append(filename)\n",
    "        sys.stderr.write(\"ended reading folder %s\\n\" % directory)\n",
    "        os.chdir('..')\n",
    "    os.chdir(cur_dir)\n",
    "    return np.array(data), np.array(labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "_DATA_PATH = './dataset'\n",
    "_CLASS_LABELS = (\"Neutral\", \"Angry\", \"Happy\", \"Sad\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "\n",
    "def extract_data(flatten):\n",
    "    data, labels = get_data(_DATA_PATH, class_labels=_CLASS_LABELS,\n",
    "                            flatten=flatten)\n",
    "    x_train, x_test, y_train, y_test = train_test_split(\n",
    "        data,\n",
    "        labels,\n",
    "        test_size=0.2,\n",
    "        random_state=42)\n",
    "    return np.array(x_train), np.array(x_test), np.array(y_train), np.array(\n",
    "        y_test), len(_CLASS_LABELS)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras import Sequential\n",
    "from keras.layers import LSTM as KERAS_LSTM, Dense, Dropout, Conv2D, Flatten, \\\n",
    "    BatchNormalization, Activation, MaxPooling2D\n",
    "from sklearn.metrics import accuracy_score, confusion_matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Model(object):\n",
    "    \"\"\"\n",
    "    Model is the abstract class which determines how a model should be.\n",
    "    Any model inheriting this class should do the following.\n",
    "\n",
    "    1.  Set the model instance variable to the corresponding model class which\n",
    "        which will provide methods `fit` and `predict`.\n",
    "\n",
    "    2.  Should implement the following abstract methods `load_model`,\n",
    "        `save_model` `train` and `evaluate`. These methods provide the\n",
    "        functionality to save the model to the disk, load the model from the\n",
    "        disk and train the model and evaluate the model to return appropriate\n",
    "        measure like accuracy, f1 score, etc.\n",
    "\n",
    "    Attributes:\n",
    "        model (Any): instance variable that holds the model.\n",
    "        save_path (str): path to save the model.\n",
    "        name (str): name of the model.\n",
    "        trained (bool): True if model has been trained, false otherwise.\n",
    "    \"\"\"\n",
    "\n",
    "    def __init__(self, save_path: str = '', name: str = 'Not Specified'):\n",
    "        \"\"\"\n",
    "        Default constructor for abstract class Model.\n",
    "\n",
    "        Args:\n",
    "            save_path(str): path to save the model to.\n",
    "            name(str): name of the model given as string.\n",
    "\n",
    "        \"\"\"\n",
    "        # Place holder for model\n",
    "        self.model = None\n",
    "        # Place holder on where to save the model\n",
    "        self.save_path = save_path\n",
    "        # Place holder for name of the model\n",
    "        self.name = name\n",
    "        # Model has been trained or not\n",
    "        self.trained = False\n",
    "\n",
    "    def train(self, x_train: np.ndarray, y_train: np.ndarray,\n",
    "              x_val: np.ndarray = None,\n",
    "              y_val: np.ndarray = None) -> None:\n",
    "        \"\"\"\n",
    "        Trains the model with the given training data.\n",
    "\n",
    "        Args:\n",
    "            x_train (numpy.ndarray): samples of training data.\n",
    "            y_train (numpy.ndarray): labels for training data.\n",
    "            x_val (numpy.ndarray): Optional, samples in the validation data.\n",
    "            y_val (numpy.ndarray): Optional, labels of the validation data.\n",
    "\n",
    "        \"\"\"\n",
    "        # This will be specific to model so should be implemented by\n",
    "        # child classes\n",
    "        raise NotImplementedError()\n",
    "\n",
    "    def predict(self, samples: np.ndarray) -> Tuple:\n",
    "        \"\"\"\n",
    "        Predict labels for given data.\n",
    "\n",
    "        Args:\n",
    "            samples (numpy.ndarray): data for which labels need to be predicted\n",
    "\n",
    "        Returns:\n",
    "            list: list of labels predicted for the data.\n",
    "\n",
    "        \"\"\"\n",
    "        results = []\n",
    "        for _, sample in enumerate(samples):\n",
    "            results.append(self.predict_one(sample))\n",
    "        return tuple(results)\n",
    "\n",
    "    def predict_one(self, sample) -> int:\n",
    "        \"\"\"\n",
    "        Predict label of a single sample. The reason this method exists is\n",
    "        because often we might want to predict label for a single sample.\n",
    "\n",
    "        Args:\n",
    "            sample (numpy.ndarray): Feature vector of the sample that we want to\n",
    "                                    predict the label for.\n",
    "\n",
    "        Returns:\n",
    "            int: returns the label for the sample.\n",
    "        \"\"\"\n",
    "        # This need to be implemented for the child models. The reason is that\n",
    "        # ML models and DL models predict the labels differently.\n",
    "        raise NotImplementedError()\n",
    "\n",
    "    def restore_model(self, load_path: str = None) -> None:\n",
    "        \"\"\"\n",
    "        Restore the weights from a saved model and load them to the model.\n",
    "\n",
    "        Args:\n",
    "            load_path (str): Optional, path to load the weights from a given path.\n",
    "\n",
    "        \"\"\"\n",
    "        to_load = load_path or self.save_path\n",
    "        if to_load is None:\n",
    "            sys.stderr.write(\n",
    "                \"Provide a path to load from or save_path of the model\\n\")\n",
    "            sys.exit(-1)\n",
    "        self.load_model(to_load)\n",
    "        self.trained = True\n",
    "\n",
    "    def load_model(self, to_load: str) -> None:\n",
    "        \"\"\"\n",
    "        Load the weights from the given saved model.\n",
    "\n",
    "        Args:\n",
    "            to_load: path containing the saved model.\n",
    "\n",
    "        \"\"\"\n",
    "        # This will be specific to model so should be implemented by\n",
    "        # child classes\n",
    "        raise NotImplementedError()\n",
    "\n",
    "    def save_model(self) -> None:\n",
    "        \"\"\"\n",
    "        Save the model to path denoted by `save_path` instance variable.\n",
    "        \"\"\"\n",
    "        # This will be specific to model so should be implemented by\n",
    "        # child classes\n",
    "        raise NotImplementedError()\n",
    "\n",
    "    def evaluate(self, x_test: np.ndarray, y_test: np.ndarray) -> None:\n",
    "        \"\"\"\n",
    "        Evaluate the current model on the given test data.\n",
    "\n",
    "        Predict the labels for test data using the model and print the relevant\n",
    "        metrics like accuracy and the confusion matrix.\n",
    "\n",
    "        Args:\n",
    "            x_test (numpy.ndarray): Numpy nD array or a list like object\n",
    "                                    containing the samples.\n",
    "            y_test (numpy.ndarray): Numpy 1D array or list like object\n",
    "                                    containing the labels for test samples.\n",
    "        \"\"\"\n",
    "        predictions = self.predict(x_test)\n",
    "        print(y_test)\n",
    "        print(predictions)\n",
    "        print('Accuracy:%.3f\\n' % accuracy_score(y_pred=predictions,\n",
    "                                                 y_true=y_test))\n",
    "        print('Confusion matrix:', confusion_matrix(y_pred=predictions,\n",
    "                                                    y_true=y_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "class DNN(Model):\n",
    "    \"\"\"\n",
    "    This class is parent class for all Deep neural network models. Any class\n",
    "    inheriting this class should implement `make_default_model` method which\n",
    "    creates a model with a set of hyper parameters.\n",
    "    \"\"\"\n",
    "\n",
    "    def __init__(self, input_shape, num_classes, **params):\n",
    "        \"\"\"\n",
    "        Constructor to initialize the deep neural network model. Takes the input\n",
    "        shape and number of classes and other parameters required for the\n",
    "        abstract class `Model` as parameters.\n",
    "\n",
    "        Args:\n",
    "            input_shape (tuple): shape of the input\n",
    "            num_classes (int): number of different classes ( labels ) in the data.\n",
    "            **params: Additional parameters required by the underlying abstract\n",
    "                class `Model`.\n",
    "\n",
    "        \"\"\"\n",
    "        super(DNN, self).__init__(**params)\n",
    "        self.input_shape = input_shape\n",
    "        self.model = Sequential()\n",
    "        self.make_default_model()\n",
    "        self.model.add(Dense(num_classes, activation='softmax'))\n",
    "        self.model.compile(loss='binary_crossentropy', optimizer='adam',\n",
    "                           metrics=['accuracy'])\n",
    "        print(self.model.summary(), file=sys.stderr)\n",
    "        self.save_path = self.save_path or self.name + '_best_model.h5'\n",
    "\n",
    "    def load_model(self, to_load):\n",
    "        \"\"\"\n",
    "        Load the model weights from the given path.\n",
    "\n",
    "        Args:\n",
    "            to_load (str): path to the saved model file in h5 format.\n",
    "\n",
    "        \"\"\"\n",
    "        try:\n",
    "            self.model.load_weights(to_load)\n",
    "        except:\n",
    "            sys.stderr.write(\"Invalid saved file provided\")\n",
    "            sys.exit(-1)\n",
    "\n",
    "    def save_model(self, path, save_format=None):\n",
    "        \"\"\"\n",
    "        Save the model weights to `save_path` provided while creating the model.\n",
    "        \"\"\"\n",
    "        # self.model.save_weights(self.save_path)\n",
    "        self.model.save(path, save_format)\n",
    "\n",
    "    def train(self, x_train, y_train, x_val=None, y_val=None, n_epochs=50):\n",
    "        \"\"\"\n",
    "        Train the model on the given training data.\n",
    "\n",
    "\n",
    "        Args:\n",
    "            x_train (numpy.ndarray): samples of training data.\n",
    "            y_train (numpy.ndarray): labels for training data.\n",
    "            x_val (numpy.ndarray): Optional, samples in the validation data.\n",
    "            y_val (numpy.ndarray): Optional, labels of the validation data.\n",
    "            n_epochs (int): Number of epochs to be trained.\n",
    "\n",
    "        \"\"\"\n",
    "        best_acc = 0\n",
    "        if x_val is None or y_val is None:\n",
    "            x_val, y_val = x_train, y_train\n",
    "        for i in range(n_epochs):\n",
    "            # Shuffle the data for each epoch in unison inspired\n",
    "            # from https://stackoverflow.com/a/4602224\n",
    "            p = np.random.permutation(len(x_train))\n",
    "            x_train = x_train[p]\n",
    "            y_train = y_train[p]\n",
    "            self.model.fit(x_train, y_train, batch_size=32, epochs=1)\n",
    "            loss, acc = self.model.evaluate(x_val, y_val)\n",
    "            if acc > best_acc:\n",
    "                best_acc = acc\n",
    "        self.trained = True\n",
    "\n",
    "    def predict_one(self, sample):\n",
    "        # if not self.trained:\n",
    "        #     sys.stderr.write(\n",
    "        #         \"Model should be trained or loaded before doing predict\\n\")\n",
    "        #     sys.exit(-1)\n",
    "        return np.argmax(self.model.predict(np.array([sample])))\n",
    "\n",
    "    def make_default_model(self) -> None:\n",
    "        \"\"\"\n",
    "        Make the model with default hyper parameters\n",
    "        \"\"\"\n",
    "        # This has to be implemented by child classes. The reason is that the\n",
    "        # hyper parameters depends on the model.\n",
    "        raise NotImplementedError()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "class CNN(DNN):\n",
    "    \"\"\"\n",
    "    This class handles CNN for speech emotion recognitions\n",
    "    \"\"\"\n",
    "\n",
    "    def __init__(self, **params):\n",
    "        params['name'] = 'CNN'\n",
    "        super(CNN, self).__init__(**params)\n",
    "\n",
    "    def make_default_model(self):\n",
    "        \"\"\"\n",
    "        Makes a CNN keras model with the default hyper parameters.\n",
    "        \"\"\"\n",
    "        self.model.add(Conv2D(8, (13, 13),\n",
    "                              input_shape=(\n",
    "                                  self.input_shape[0], self.input_shape[1], 1)))\n",
    "        self.model.add(BatchNormalization(axis=-1))\n",
    "        self.model.add(Activation('relu'))\n",
    "        self.model.add(Conv2D(8, (13, 13)))\n",
    "        self.model.add(BatchNormalization(axis=-1))\n",
    "        self.model.add(Activation('relu'))\n",
    "        self.model.add(MaxPooling2D(pool_size=(2, 1)))\n",
    "        self.model.add(Conv2D(8, (13, 13)))\n",
    "        self.model.add(BatchNormalization(axis=-1))\n",
    "        self.model.add(Activation('relu'))\n",
    "        self.model.add(Conv2D(8, (2, 2)))\n",
    "        self.model.add(BatchNormalization(axis=-1))\n",
    "        self.model.add(Activation('relu'))\n",
    "        self.model.add(MaxPooling2D(pool_size=(2, 1)))\n",
    "        self.model.add(Flatten())\n",
    "        self.model.add(Dense(64))\n",
    "        self.model.add(BatchNormalization())\n",
    "        self.model.add(Activation('relu'))\n",
    "        self.model.add(Dropout(0.2))\n",
    "\n",
    "\n",
    "class LSTM(DNN):\n",
    "    \"\"\"\n",
    "    This class handles CNN for speech emotion recognitions\n",
    "    \"\"\"\n",
    "\n",
    "    def __init__(self, **params):\n",
    "        params['name'] = 'LSTM'\n",
    "        super(LSTM, self).__init__(**params)\n",
    "\n",
    "    def make_default_model(self):\n",
    "        \"\"\"\n",
    "        Makes the LSTM model with keras with the default hyper parameters.\n",
    "        \"\"\"\n",
    "        self.model.add(\n",
    "            KERAS_LSTM(128,\n",
    "                       input_shape=(self.input_shape[0], self.input_shape[1])))\n",
    "        self.model.add(Dropout(0.5))\n",
    "        self.model.add(Dense(32, activation='relu'))\n",
    "        self.model.add(Dense(16, activation='tanh'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.utils import to_categorical"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "curdir: /home/yuvraj/projects/Speak-Secure\n",
      "started reading folder Neutral\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "29215\n",
      "32000\n",
      "32000\n",
      "49240\n",
      "32000\n",
      "32000\n",
      "38049\n",
      "32000\n",
      "32000\n",
      "61105\n",
      "32000\n",
      "32000\n",
      "32799\n",
      "32000\n",
      "32000\n",
      "26538\n",
      "32000\n",
      "32000\n",
      "42766\n",
      "32000\n",
      "32000\n",
      "38227\n",
      "32000\n",
      "32000\n",
      "30587\n",
      "32000\n",
      "32000\n",
      "58312\n",
      "32000\n",
      "32000\n",
      "50688\n",
      "32000\n",
      "32000\n",
      "41162\n",
      "32000\n",
      "32000\n",
      "22893\n",
      "32000\n",
      "32000\n",
      "46391\n",
      "32000\n",
      "32000\n",
      "52045\n",
      "32000\n",
      "32000\n",
      "29375\n",
      "32000\n",
      "32000\n",
      "32532\n",
      "32000\n",
      "32000\n",
      "44644\n",
      "32000\n",
      "32000\n",
      "57935\n",
      "32000\n",
      "32000\n",
      "28915\n",
      "32000\n",
      "32000\n",
      "24250\n",
      "32000\n",
      "32000\n",
      "28610\n",
      "32000\n",
      "32000\n",
      "36103\n",
      "32000\n",
      "32000\n",
      "32113\n",
      "32000\n",
      "32000\n",
      "43552\n",
      "32000\n",
      "32000\n",
      "27170\n",
      "32000\n",
      "32000\n",
      "24933\n",
      "32000\n",
      "32000\n",
      "48888\n",
      "32000\n",
      "32000\n",
      "43425\n",
      "32000\n",
      "32000\n",
      "26921\n",
      "32000\n",
      "32000\n",
      "49013\n",
      "32000\n",
      "32000\n",
      "24545\n",
      "32000\n",
      "32000\n",
      "28650\n",
      "32000\n",
      "32000\n",
      "25780\n",
      "32000\n",
      "32000\n",
      "37664\n",
      "32000\n",
      "32000\n",
      "34121\n",
      "32000\n",
      "32000\n",
      "35768\n",
      "32000\n",
      "32000\n",
      "24981\n",
      "32000\n",
      "32000\n",
      "45660\n",
      "32000\n",
      "32000\n",
      "26944\n",
      "32000\n",
      "32000\n",
      "26052\n",
      "32000\n",
      "32000\n",
      "47125\n",
      "32000\n",
      "32000\n",
      "51004\n",
      "32000\n",
      "32000\n",
      "62387\n",
      "32000\n",
      "32000\n",
      "27711\n",
      "32000\n",
      "32000\n",
      "39449\n",
      "32000\n",
      "32000\n",
      "37716\n",
      "32000\n",
      "32000\n",
      "47713\n",
      "32000\n",
      "32000\n",
      "33343\n",
      "32000\n",
      "32000\n",
      "26182\n",
      "32000\n",
      "32000\n",
      "25845\n",
      "32000\n",
      "32000\n",
      "57072\n",
      "32000\n",
      "32000\n",
      "35523\n",
      "32000\n",
      "32000\n",
      "43204\n",
      "32000\n",
      "32000\n",
      "49935\n",
      "32000\n",
      "32000\n",
      "35244\n",
      "32000\n",
      "32000\n",
      "39238\n",
      "32000\n",
      "32000\n",
      "30099\n",
      "32000\n",
      "32000\n",
      "28351\n",
      "32000\n",
      "32000\n",
      "32927\n",
      "32000\n",
      "32000\n",
      "28281\n",
      "32000\n",
      "32000\n",
      "50974\n",
      "32000\n",
      "32000\n",
      "41417\n",
      "32000\n",
      "32000\n",
      "35412\n",
      "32000\n",
      "32000\n",
      "35483\n",
      "32000\n",
      "32000\n",
      "38708\n",
      "32000\n",
      "32000\n",
      "51207\n",
      "32000\n",
      "32000\n",
      "25822\n",
      "32000\n",
      "32000\n",
      "53128\n",
      "32000\n",
      "32000\n",
      "23037\n",
      "32000\n",
      "32000\n",
      "27551\n",
      "32000\n",
      "32000\n",
      "33537\n",
      "32000\n",
      "32000\n",
      "38896\n",
      "32000\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/tmp/ipykernel_46399/284012828.py:13: WavFileWarning: Chunk (non-data) not understood, skipping it.\n",
      "  fs, signal = wav.read(file_path)\n",
      "ended reading folder Neutral\n",
      "started reading folder Angry\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "32000\n",
      "26567\n",
      "32000\n",
      "32000\n",
      "59547\n",
      "32000\n",
      "32000\n",
      "43181\n",
      "32000\n",
      "32000\n",
      "44580\n",
      "32000\n",
      "32000\n",
      "28232\n",
      "32000\n",
      "32000\n",
      "37885\n",
      "32000\n",
      "32000\n",
      "23969\n",
      "32000\n",
      "32000\n",
      "60812\n",
      "32000\n",
      "32000\n",
      "42863\n",
      "32000\n",
      "32000\n",
      "44866\n",
      "32000\n",
      "32000\n",
      "28879\n",
      "32000\n",
      "32000\n",
      "35209\n",
      "32000\n",
      "32000\n",
      "56725\n",
      "32000\n",
      "32000\n",
      "31105\n",
      "32000\n",
      "32000\n",
      "26485\n",
      "32000\n",
      "32000\n",
      "57790\n",
      "32000\n",
      "32000\n",
      "52983\n",
      "32000\n",
      "32000\n",
      "59040\n",
      "32000\n",
      "32000\n",
      "50649\n",
      "32000\n",
      "32000\n",
      "38394\n",
      "32000\n",
      "32000\n",
      "44331\n",
      "32000\n",
      "32000\n",
      "38626\n",
      "32000\n",
      "32000\n",
      "38577\n",
      "32000\n",
      "32000\n",
      "57344\n",
      "32000\n",
      "32000\n",
      "40161\n",
      "32000\n",
      "32000\n",
      "41758\n",
      "32000\n",
      "32000\n",
      "39072\n",
      "32000\n",
      "32000\n",
      "39011\n",
      "32000\n",
      "32000\n",
      "41957\n",
      "32000\n",
      "32000\n",
      "47396\n",
      "32000\n",
      "32000\n",
      "50945\n",
      "32000\n",
      "32000\n",
      "36035\n",
      "32000\n",
      "32000\n",
      "26205\n",
      "32000\n",
      "32000\n",
      "46300\n",
      "32000\n",
      "32000\n",
      "35198\n",
      "32000\n",
      "32000\n",
      "37109\n",
      "32000\n",
      "32000\n",
      "25242\n",
      "32000\n",
      "32000\n",
      "35814\n",
      "32000\n",
      "32000\n",
      "62793\n",
      "32000\n",
      "32000\n",
      "44540\n",
      "32000\n",
      "32000\n",
      "27056\n",
      "32000\n",
      "32000\n",
      "28982\n",
      "32000\n",
      "32000\n",
      "46153\n",
      "32000\n",
      "32000\n",
      "64109\n",
      "32000\n",
      "32000\n",
      "31927\n",
      "32000\n",
      "32000\n",
      "59121\n",
      "32000\n",
      "32000\n",
      "31802\n",
      "32000\n",
      "32000\n",
      "28979\n",
      "32000\n",
      "32000\n",
      "41973\n",
      "32000\n",
      "32000\n",
      "49576\n",
      "32000\n",
      "32000\n",
      "42398\n",
      "32000\n",
      "32000\n",
      "42339\n",
      "32000\n",
      "32000\n",
      "35509\n",
      "32000\n",
      "32000\n",
      "42933\n",
      "32000\n",
      "32000\n",
      "26704\n",
      "32000\n",
      "32000\n",
      "24119\n",
      "32000\n",
      "32000\n",
      "53807\n",
      "32000\n",
      "32000\n",
      "38815\n",
      "32000\n",
      "32000\n",
      "42485\n",
      "32000\n",
      "32000\n",
      "71768\n",
      "32000\n",
      "32000\n",
      "33016\n",
      "32000\n",
      "32000\n",
      "31559\n",
      "32000\n",
      "32000\n",
      "37796\n",
      "32000\n",
      "32000\n",
      "29389\n",
      "32000\n",
      "32000\n",
      "30045\n",
      "32000\n",
      "32000\n",
      "32204\n",
      "32000\n",
      "32000\n",
      "34877\n",
      "32000\n",
      "32000\n",
      "37324\n",
      "32000\n",
      "32000\n",
      "31126\n",
      "32000\n",
      "32000\n",
      "34663\n",
      "32000\n",
      "32000\n",
      "37741\n",
      "32000\n",
      "32000\n",
      "42494\n",
      "32000\n",
      "32000\n",
      "45654\n",
      "32000\n",
      "32000\n",
      "54239\n",
      "32000\n",
      "32000\n",
      "25805\n",
      "32000\n",
      "32000\n",
      "44241\n",
      "32000\n",
      "32000\n",
      "28915\n",
      "32000\n",
      "32000\n",
      "36122\n",
      "32000\n",
      "32000\n",
      "48346\n",
      "32000\n",
      "32000\n",
      "40360\n",
      "32000\n",
      "32000\n",
      "64612\n",
      "32000\n",
      "32000\n",
      "47333\n",
      "32000\n",
      "32000\n",
      "48865\n",
      "32000\n",
      "32000\n",
      "50954\n",
      "32000\n",
      "32000\n",
      "33978\n",
      "32000\n",
      "32000\n",
      "35796\n",
      "32000\n",
      "32000\n",
      "28321\n",
      "32000\n",
      "32000\n",
      "23453\n",
      "32000\n",
      "32000\n",
      "35306\n",
      "32000\n",
      "32000\n",
      "59857\n",
      "32000\n",
      "32000\n",
      "61559\n",
      "32000\n",
      "32000\n",
      "34725\n",
      "32000\n",
      "32000\n",
      "27224\n",
      "32000\n",
      "32000\n",
      "42102\n",
      "32000\n",
      "32000\n",
      "46173\n",
      "32000\n",
      "32000\n",
      "42587\n",
      "32000\n",
      "32000\n",
      "55052\n",
      "32000\n",
      "32000\n",
      "29875\n",
      "32000\n",
      "32000\n",
      "43970\n",
      "32000\n",
      "32000\n",
      "40991\n",
      "32000\n",
      "32000\n",
      "29131\n",
      "32000\n",
      "32000\n",
      "43592\n",
      "32000\n",
      "32000\n",
      "54403\n",
      "32000\n",
      "32000\n",
      "50166\n",
      "32000\n",
      "32000\n",
      "55764\n",
      "32000\n",
      "32000\n",
      "27789\n",
      "32000\n",
      "32000\n",
      "51416\n",
      "32000\n",
      "32000\n",
      "33401\n",
      "32000\n",
      "32000\n",
      "83416\n",
      "32000\n",
      "32000\n",
      "60033\n",
      "32000\n",
      "32000\n",
      "36677\n",
      "32000\n",
      "32000\n",
      "65300\n",
      "32000\n",
      "32000\n",
      "40789\n",
      "32000\n",
      "32000\n",
      "30716\n",
      "32000\n",
      "32000\n",
      "33643\n",
      "32000\n",
      "32000\n",
      "46867\n",
      "32000\n",
      "32000\n",
      "32706\n",
      "32000\n",
      "32000\n",
      "51978\n",
      "32000\n",
      "32000\n",
      "45035\n",
      "32000\n",
      "32000\n",
      "65037\n",
      "32000\n",
      "32000\n",
      "46868\n",
      "32000\n",
      "32000\n",
      "38109\n",
      "32000\n",
      "32000\n",
      "31575\n",
      "32000\n",
      "32000\n",
      "45889\n",
      "32000\n",
      "32000\n",
      "62179\n",
      "32000\n",
      "32000\n",
      "33482\n",
      "32000\n",
      "32000\n",
      "29301\n",
      "32000\n",
      "32000\n",
      "50950\n",
      "32000\n",
      "32000\n",
      "41941\n",
      "32000\n",
      "32000\n",
      "42140\n",
      "32000\n",
      "32000\n",
      "60960\n",
      "32000\n",
      "32000\n",
      "52936\n",
      "32000\n",
      "32000\n",
      "34519\n",
      "32000\n",
      "32000\n",
      "56537\n",
      "32000\n",
      "32000\n",
      "44269\n",
      "32000\n",
      "32000\n",
      "30176\n",
      "32000\n",
      "32000\n",
      "61848\n",
      "32000\n",
      "32000\n",
      "51139\n",
      "32000\n",
      "32000\n",
      "27149\n",
      "32000\n",
      "32000\n",
      "39417\n",
      "32000\n",
      "32000\n",
      "52864\n",
      "32000\n",
      "32000\n",
      "34141\n",
      "32000\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "ended reading folder Angry\n",
      "started reading folder Happy\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "32000\n",
      "24384\n",
      "32000\n",
      "32000\n",
      "39711\n",
      "32000\n",
      "32000\n",
      "37795\n",
      "32000\n",
      "32000\n",
      "23702\n",
      "32000\n",
      "32000\n",
      "25912\n",
      "32000\n",
      "32000\n",
      "40654\n",
      "32000\n",
      "32000\n",
      "31427\n",
      "32000\n",
      "32000\n",
      "42396\n",
      "32000\n",
      "32000\n",
      "58519\n",
      "32000\n",
      "32000\n",
      "30095\n",
      "32000\n",
      "32000\n",
      "41336\n",
      "32000\n",
      "32000\n",
      "53000\n",
      "32000\n",
      "32000\n",
      "62895\n",
      "32000\n",
      "32000\n",
      "57857\n",
      "32000\n",
      "32000\n",
      "48585\n",
      "32000\n",
      "32000\n",
      "39215\n",
      "32000\n",
      "32000\n",
      "26907\n",
      "32000\n",
      "32000\n",
      "42128\n",
      "32000\n",
      "32000\n",
      "24595\n",
      "32000\n",
      "32000\n",
      "38829\n",
      "32000\n",
      "32000\n",
      "28565\n",
      "32000\n",
      "32000\n",
      "32702\n",
      "32000\n",
      "32000\n",
      "50059\n",
      "32000\n",
      "32000\n",
      "47396\n",
      "32000\n",
      "32000\n",
      "33291\n",
      "32000\n",
      "32000\n",
      "51939\n",
      "32000\n",
      "32000\n",
      "49839\n",
      "32000\n",
      "32000\n",
      "31409\n",
      "32000\n",
      "32000\n",
      "28659\n",
      "32000\n",
      "32000\n",
      "30372\n",
      "32000\n",
      "32000\n",
      "40879\n",
      "32000\n",
      "32000\n",
      "37940\n",
      "32000\n",
      "32000\n",
      "50267\n",
      "32000\n",
      "32000\n",
      "40252\n",
      "32000\n",
      "32000\n",
      "32100\n",
      "32000\n",
      "32000\n",
      "55071\n",
      "32000\n",
      "32000\n",
      "43139\n",
      "32000\n",
      "32000\n",
      "29818\n",
      "32000\n",
      "32000\n",
      "56696\n",
      "32000\n",
      "32000\n",
      "40964\n",
      "32000\n",
      "32000\n",
      "28533\n",
      "32000\n",
      "32000\n",
      "56029\n",
      "32000\n",
      "32000\n",
      "30964\n",
      "32000\n",
      "32000\n",
      "57798\n",
      "32000\n",
      "32000\n",
      "28810\n",
      "32000\n",
      "32000\n",
      "38929\n",
      "32000\n",
      "32000\n",
      "33542\n",
      "32000\n",
      "32000\n",
      "45395\n",
      "32000\n",
      "32000\n",
      "36796\n",
      "32000\n",
      "32000\n",
      "45981\n",
      "32000\n",
      "32000\n",
      "34537\n",
      "32000\n",
      "32000\n",
      "61277\n",
      "32000\n",
      "32000\n",
      "32199\n",
      "32000\n",
      "32000\n",
      "26026\n",
      "32000\n",
      "32000\n",
      "62823\n",
      "32000\n",
      "32000\n",
      "29298\n",
      "32000\n",
      "32000\n",
      "44338\n",
      "32000\n",
      "32000\n",
      "49579\n",
      "32000\n",
      "32000\n",
      "39198\n",
      "32000\n",
      "32000\n",
      "35388\n",
      "32000\n",
      "32000\n",
      "40472\n",
      "32000\n",
      "32000\n",
      "33195\n",
      "32000\n",
      "32000\n",
      "90773\n",
      "32000\n",
      "32000\n",
      "79620\n",
      "32000\n",
      "32000\n",
      "60102\n",
      "32000\n",
      "32000\n",
      "41721\n",
      "32000\n",
      "32000\n",
      "34198\n",
      "32000\n",
      "32000\n",
      "95941\n",
      "32000\n",
      "32000\n",
      "80462\n",
      "32000\n",
      "32000\n",
      "54583\n",
      "32000\n",
      "32000\n",
      "78389\n",
      "32000\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "ended reading folder Happy\n",
      "started reading folder Sad\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "32000\n",
      "108023\n",
      "32000\n",
      "32000\n",
      "52216\n",
      "32000\n",
      "32000\n",
      "35071\n",
      "32000\n",
      "32000\n",
      "64252\n",
      "32000\n",
      "32000\n",
      "87957\n",
      "32000\n",
      "32000\n",
      "37467\n",
      "32000\n",
      "32000\n",
      "36267\n",
      "32000\n",
      "32000\n",
      "35609\n",
      "32000\n",
      "32000\n",
      "62272\n",
      "32000\n",
      "32000\n",
      "84789\n",
      "32000\n",
      "32000\n",
      "95610\n",
      "32000\n",
      "32000\n",
      "85457\n",
      "32000\n",
      "32000\n",
      "31908\n",
      "32000\n",
      "32000\n",
      "54251\n",
      "32000\n",
      "32000\n",
      "33017\n",
      "32000\n",
      "32000\n",
      "56010\n",
      "32000\n",
      "32000\n",
      "143652\n",
      "32000\n",
      "32000\n",
      "55965\n",
      "32000\n",
      "32000\n",
      "122871\n",
      "32000\n",
      "32000\n",
      "71222\n",
      "32000\n",
      "32000\n",
      "89943\n",
      "32000\n",
      "32000\n",
      "60243\n",
      "32000\n",
      "32000\n",
      "27771\n",
      "32000\n",
      "32000\n",
      "33032\n",
      "32000\n",
      "32000\n",
      "74605\n",
      "32000\n",
      "32000\n",
      "62538\n",
      "32000\n",
      "32000\n",
      "62233\n",
      "32000\n",
      "32000\n",
      "65492\n",
      "32000\n",
      "32000\n",
      "31293\n",
      "32000\n",
      "32000\n",
      "57668\n",
      "32000\n",
      "32000\n",
      "94480\n",
      "32000\n",
      "32000\n",
      "29820\n",
      "32000\n",
      "32000\n",
      "34678\n",
      "32000\n",
      "32000\n",
      "48745\n",
      "32000\n",
      "32000\n",
      "59672\n",
      "32000\n",
      "32000\n",
      "62947\n",
      "32000\n",
      "32000\n",
      "82030\n",
      "32000\n",
      "32000\n",
      "46956\n",
      "32000\n",
      "32000\n",
      "85764\n",
      "32000\n",
      "32000\n",
      "63154\n",
      "32000\n",
      "32000\n",
      "99447\n",
      "32000\n",
      "32000\n",
      "50651\n",
      "32000\n",
      "32000\n",
      "45640\n",
      "32000\n",
      "32000\n",
      "86622\n",
      "32000\n",
      "32000\n",
      "59800\n",
      "32000\n",
      "32000\n",
      "63927\n",
      "32000\n",
      "32000\n",
      "64377\n",
      "32000\n",
      "32000\n",
      "58475\n",
      "32000\n",
      "32000\n",
      "55404\n",
      "32000\n",
      "32000\n",
      "53343\n",
      "32000\n",
      "32000\n",
      "61395\n",
      "32000\n",
      "32000\n",
      "108608\n",
      "32000\n",
      "32000\n",
      "70044\n",
      "32000\n",
      "32000\n",
      "(271, 198, 39)\n",
      "(271, 198, 39, 1)\n",
      "Model: \"sequential_3\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " conv2d_12 (Conv2D)          (None, 186, 27, 8)        1360      \n",
      "                                                                 \n",
      " batch_normalization_15 (Ba  (None, 186, 27, 8)        32        \n",
      " tchNormalization)                                               \n",
      "                                                                 \n",
      " activation_15 (Activation)  (None, 186, 27, 8)        0         \n",
      "                                                                 \n",
      " conv2d_13 (Conv2D)          (None, 174, 15, 8)        10824     \n",
      "                                                                 \n",
      " batch_normalization_16 (Ba  (None, 174, 15, 8)        32        \n",
      " tchNormalization)                                               \n",
      "                                                                 \n",
      " activation_16 (Activation)  (None, 174, 15, 8)        0         \n",
      "                                                                 \n",
      " max_pooling2d_6 (MaxPoolin  (None, 87, 15, 8)         0         \n",
      " g2D)                                                            \n",
      "                                                                 \n",
      " conv2d_14 (Conv2D)          (None, 75, 3, 8)          10824     \n",
      "                                                                 \n",
      " batch_normalization_17 (Ba  (None, 75, 3, 8)          32        \n",
      " tchNormalization)                                               \n",
      "                                                                 \n",
      " activation_17 (Activation)  (None, 75, 3, 8)          0         \n",
      "                                                                 \n",
      " conv2d_15 (Conv2D)          (None, 74, 2, 8)          264       \n",
      "                                                                 \n",
      " batch_normalization_18 (Ba  (None, 74, 2, 8)          32        \n",
      " tchNormalization)                                               \n",
      "                                                                 \n",
      " activation_18 (Activation)  (None, 74, 2, 8)          0         \n",
      "                                                                 \n",
      " max_pooling2d_7 (MaxPoolin  (None, 37, 2, 8)          0         \n",
      " g2D)                                                            \n",
      "                                                                 \n",
      " flatten_3 (Flatten)         (None, 592)               0         \n",
      "                                                                 \n",
      " dense_6 (Dense)             (None, 64)                37952     \n",
      "                                                                 \n",
      " batch_normalization_19 (Ba  (None, 64)                256       \n",
      " tchNormalization)                                               \n",
      "                                                                 \n",
      " activation_19 (Activation)  (None, 64)                0         \n",
      "                                                                 \n",
      " dropout_3 (Dropout)         (None, 64)                0         \n",
      "                                                                 \n",
      " dense_7 (Dense)             (None, 4)                 260       \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 61868 (241.67 KB)\n",
      "Trainable params: 61676 (240.92 KB)\n",
      "Non-trainable params: 192 (768.00 Byte)\n",
      "_________________________________________________________________\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "ended reading folder Sad\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "x_train, x_test, y_train, y_test, num_labels = extract_data(\n",
    "    flatten=False)\n",
    "y_train = to_categorical(y_train)\n",
    "y_test_train = to_categorical(y_test)\n",
    "print(x_train.shape)\n",
    "in_shape = x_train[0].shape\n",
    "x_train = x_train.reshape(x_train.shape[0], in_shape[0], in_shape[1], 1)\n",
    "print(x_train.shape)\n",
    "x_test = x_test.reshape(x_test.shape[0], in_shape[0], in_shape[1], 1)\n",
    "model = CNN(input_shape=x_train[0].shape,\n",
    "            num_classes=num_labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {},
   "outputs": [],
   "source": [
    "features2 = np.expand_dims(features, axis=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(198, 39, 1)"
      ]
     },
     "execution_count": 89,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "features2.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.load_model('./sentiment_models/best_model_CNN_13.h5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/yuvraj/anaconda3/envs/voice/lib/python3.11/site-packages/keras/src/engine/training.py:3000: UserWarning: You are saving your model as an HDF5 file via `model.save()`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')`.\n",
      "  saving_api.save_model(\n"
     ]
    }
   ],
   "source": [
    "model.save_model(\"./sentiment_models/sentiment_classifier.h5\", save_format='h5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 17ms/step\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "'Sad'"
      ]
     },
     "execution_count": 91,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "_CLASS_LABELS[model.predict_one(features2)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/tmp/ipykernel_30637/2756558424.py:1: UserWarning: You are saving your model as an HDF5 file via `model.save()`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')`.\n",
      "  tf.keras.saving.save_model(\n"
     ]
    },
    {
     "ename": "AttributeError",
     "evalue": "'CNN' object has no attribute 'outputs'",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mAttributeError\u001b[0m                            Traceback (most recent call last)",
      "\u001b[1;32m/home/yuvraj/projects/Speak-Secure/sentiment.ipynb Cell 21\u001b[0m line \u001b[0;36m1\n\u001b[0;32m----> <a href='vscode-notebook-cell:/home/yuvraj/projects/Speak-Secure/sentiment.ipynb#W5sZmlsZQ%3D%3D?line=0'>1</a>\u001b[0m tf\u001b[39m.\u001b[39mkeras\u001b[39m.\u001b[39msaving\u001b[39m.\u001b[39msave_model(\n\u001b[1;32m      <a href='vscode-notebook-cell:/home/yuvraj/projects/Speak-Secure/sentiment.ipynb#W5sZmlsZQ%3D%3D?line=1'>2</a>\u001b[0m     model, \u001b[39m\"\u001b[39m\u001b[39m./sentiment_models/sentiment_classifier.h5\u001b[39m\u001b[39m\"\u001b[39m, save_format\u001b[39m=\u001b[39m\u001b[39m\"\u001b[39m\u001b[39mh5\u001b[39m\u001b[39m\"\u001b[39m)\n",
      "File \u001b[0;32m~/anaconda3/envs/voice/lib/python3.11/site-packages/keras/src/saving/saving_api.py:149\u001b[0m, in \u001b[0;36msave_model\u001b[0;34m(model, filepath, overwrite, save_format, **kwargs)\u001b[0m\n\u001b[1;32m    146\u001b[0m     saving_lib\u001b[39m.\u001b[39msave_model(model, filepath)\n\u001b[1;32m    147\u001b[0m \u001b[39melse\u001b[39;00m:\n\u001b[1;32m    148\u001b[0m     \u001b[39m# Legacy case\u001b[39;00m\n\u001b[0;32m--> 149\u001b[0m     \u001b[39mreturn\u001b[39;00m legacy_sm_saving_lib\u001b[39m.\u001b[39msave_model(\n\u001b[1;32m    150\u001b[0m         model,\n\u001b[1;32m    151\u001b[0m         filepath,\n\u001b[1;32m    152\u001b[0m         overwrite\u001b[39m=\u001b[39moverwrite,\n\u001b[1;32m    153\u001b[0m         save_format\u001b[39m=\u001b[39msave_format,\n\u001b[1;32m    154\u001b[0m         \u001b[39m*\u001b[39m\u001b[39m*\u001b[39mkwargs,\n\u001b[1;32m    155\u001b[0m     )\n",
      "File \u001b[0;32m~/anaconda3/envs/voice/lib/python3.11/site-packages/keras/src/utils/traceback_utils.py:70\u001b[0m, in \u001b[0;36mfilter_traceback.<locals>.error_handler\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m     67\u001b[0m     filtered_tb \u001b[39m=\u001b[39m _process_traceback_frames(e\u001b[39m.\u001b[39m__traceback__)\n\u001b[1;32m     68\u001b[0m     \u001b[39m# To get the full stack trace, call:\u001b[39;00m\n\u001b[1;32m     69\u001b[0m     \u001b[39m# `tf.debugging.disable_traceback_filtering()`\u001b[39;00m\n\u001b[0;32m---> 70\u001b[0m     \u001b[39mraise\u001b[39;00m e\u001b[39m.\u001b[39mwith_traceback(filtered_tb) \u001b[39mfrom\u001b[39;00m \u001b[39mNone\u001b[39;00m\n\u001b[1;32m     71\u001b[0m \u001b[39mfinally\u001b[39;00m:\n\u001b[1;32m     72\u001b[0m     \u001b[39mdel\u001b[39;00m filtered_tb\n",
      "File \u001b[0;32m~/anaconda3/envs/voice/lib/python3.11/site-packages/keras/src/saving/legacy/saving_utils.py:351\u001b[0m, in \u001b[0;36mtry_build_compiled_arguments\u001b[0;34m(model)\u001b[0m\n\u001b[1;32m    348\u001b[0m \u001b[39mdef\u001b[39;00m \u001b[39mtry_build_compiled_arguments\u001b[39m(model):\n\u001b[1;32m    349\u001b[0m     \u001b[39mif\u001b[39;00m (\n\u001b[1;32m    350\u001b[0m         \u001b[39mnot\u001b[39;00m version_utils\u001b[39m.\u001b[39mis_v1_layer_or_model(model)\n\u001b[0;32m--> 351\u001b[0m         \u001b[39mand\u001b[39;00m model\u001b[39m.\u001b[39moutputs \u001b[39mis\u001b[39;00m \u001b[39mnot\u001b[39;00m \u001b[39mNone\u001b[39;00m\n\u001b[1;32m    352\u001b[0m     ):\n\u001b[1;32m    353\u001b[0m         \u001b[39mtry\u001b[39;00m:\n\u001b[1;32m    354\u001b[0m             \u001b[39mif\u001b[39;00m \u001b[39mnot\u001b[39;00m model\u001b[39m.\u001b[39mcompiled_loss\u001b[39m.\u001b[39mbuilt:\n",
      "\u001b[0;31mAttributeError\u001b[0m: 'CNN' object has no attribute 'outputs'"
     ]
    }
   ],
   "source": [
    "tf.keras.saving.save_model(\n",
    "    model, \"./sentiment_models/sentiment_classifier.h5\", save_format=\"h5\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.11.5"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}