audioAnalytics-3.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "#load data\n",
    "#split data into train & test sets\n",
    "#build network architecture\n",
    "#compile network\n",
    "#train network\n",
    "\n",
    "import json\n",
    "import numpy as np\n",
    "from sklearn.model_selection import train_test_split\n",
    "import tensorflow.keras as keras\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "DATASET_PATH=\"data.json\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "def load_data(dataset_path):\n",
    "    with open(dataset_path, \"r\") as fp:\n",
    "        data = json.load(fp)\n",
    "        \n",
    "    #convert lists into numpy arrays - labels and mfcc\n",
    "    inputs = np.array(data[\"mfcc\"])\n",
    "    targets = np.array(data[\"labels\"])\n",
    "    \n",
    "    return inputs, targets\n",
    "\n",
    "def plot_history(history):\n",
    "    fig, axs = plt.subplots(2)\n",
    "    \n",
    "    #create accuracy subplot\n",
    "    axs[0].plot(history.history[\"accuracy\"], label=\"train accuracy\")\n",
    "    axs[0].plot(history.history[\"val_accuracy\"], label=\"test accuracy\")\n",
    "    axs[0].set_ylabel(\"Accuracy\")\n",
    "    axs[0].legend(loc=\"lower right\")\n",
    "    axs[0].set_title(\"Accuracy Evaluation\")\n",
    "\n",
    "    \n",
    "    #create error subplot\n",
    "    axs[1].plot(history.history[\"loss\"], label=\"train error\")\n",
    "    axs[1].plot(history.history[\"val_loss\"], label=\"test error\")\n",
    "    axs[1].set_ylabel(\"Test Error\")\n",
    "    axs[1].set_xlabel(\"Epoch\")\n",
    "    axs[1].legend(loc=\"upper right\")\n",
    "    axs[1].set_title(\"Error Evaluation\")\n",
    "    \n",
    "    #plot accuracy \n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_8\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "flatten_8 (Flatten)          (None, 1690)              0         \n",
      "_________________________________________________________________\n",
      "dense_32 (Dense)             (None, 512)               865792    \n",
      "_________________________________________________________________\n",
      "dropout (Dropout)            (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "dense_33 (Dense)             (None, 256)               131328    \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 256)               0         \n",
      "_________________________________________________________________\n",
      "dense_34 (Dense)             (None, 64)                16448     \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 64)                0         \n",
      "_________________________________________________________________\n",
      "dense_35 (Dense)             (None, 10)                650       \n",
      "=================================================================\n",
      "Total params: 1,014,218\n",
      "Trainable params: 1,014,218\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "Train on 70 samples, validate on 30 samples\n",
      "Epoch 1/50\n",
      "70/70 [==============================] - 2s 31ms/sample - loss: 82.6035 - accuracy: 0.1286 - val_loss: 25.2882 - val_accuracy: 0.1000\n",
      "Epoch 2/50\n",
      "70/70 [==============================] - 0s 3ms/sample - loss: 74.5572 - accuracy: 0.1429 - val_loss: 18.4697 - val_accuracy: 0.1000\n",
      "Epoch 3/50\n",
      "70/70 [==============================] - 0s 3ms/sample - loss: 59.2335 - accuracy: 0.1571 - val_loss: 15.6422 - val_accuracy: 0.2333\n",
      "Epoch 4/50\n",
      "70/70 [==============================] - 0s 3ms/sample - loss: 53.7933 - accuracy: 0.1143 - val_loss: 14.2253 - val_accuracy: 0.2667\n",
      "Epoch 5/50\n",
      "70/70 [==============================] - 0s 3ms/sample - loss: 54.1300 - accuracy: 0.1429 - val_loss: 12.8287 - val_accuracy: 0.2333\n",
      "Epoch 6/50\n",
      "70/70 [==============================] - 0s 3ms/sample - loss: 44.9315 - accuracy: 0.1143 - val_loss: 11.1148 - val_accuracy: 0.2667\n",
      "Epoch 7/50\n",
      "70/70 [==============================] - 0s 5ms/sample - loss: 37.0960 - accuracy: 0.1571 - val_loss: 9.2581 - val_accuracy: 0.3667\n",
      "Epoch 8/50\n",
      "70/70 [==============================] - 0s 4ms/sample - loss: 30.9887 - accuracy: 0.2429 - val_loss: 8.9139 - val_accuracy: 0.3000\n",
      "Epoch 9/50\n",
      "70/70 [==============================] - 0s 4ms/sample - loss: 35.7128 - accuracy: 0.2143 - val_loss: 8.3460 - val_accuracy: 0.3000\n",
      "Epoch 10/50\n",
      "70/70 [==============================] - 0s 4ms/sample - loss: 31.2171 - accuracy: 0.2571 - val_loss: 7.5142 - val_accuracy: 0.3000\n",
      "Epoch 11/50\n",
      "70/70 [==============================] - 0s 5ms/sample - loss: 29.4063 - accuracy: 0.2000 - val_loss: 7.1128 - val_accuracy: 0.3000\n",
      "Epoch 12/50\n",
      "70/70 [==============================] - 0s 4ms/sample - loss: 27.2529 - accuracy: 0.2000 - val_loss: 6.5997 - val_accuracy: 0.4000\n",
      "Epoch 13/50\n",
      "70/70 [==============================] - 0s 4ms/sample - loss: 22.9935 - accuracy: 0.3286 - val_loss: 6.3886 - val_accuracy: 0.3667\n",
      "Epoch 14/50\n",
      "70/70 [==============================] - 0s 4ms/sample - loss: 24.7709 - accuracy: 0.2571 - val_loss: 6.6272 - val_accuracy: 0.3667\n",
      "Epoch 15/50\n",
      "70/70 [==============================] - 0s 4ms/sample - loss: 24.8204 - accuracy: 0.3286 - val_loss: 6.5684 - val_accuracy: 0.3667\n",
      "Epoch 16/50\n",
      "70/70 [==============================] - 0s 4ms/sample - loss: 20.8030 - accuracy: 0.2571 - val_loss: 6.3558 - val_accuracy: 0.4000\n",
      "Epoch 17/50\n",
      "70/70 [==============================] - 0s 3ms/sample - loss: 25.1035 - accuracy: 0.2429 - val_loss: 6.0187 - val_accuracy: 0.5000\n",
      "Epoch 18/50\n",
      "70/70 [==============================] - 0s 4ms/sample - loss: 20.3003 - accuracy: 0.2857 - val_loss: 5.4794 - val_accuracy: 0.5333\n",
      "Epoch 19/50\n",
      "70/70 [==============================] - 1s 9ms/sample - loss: 20.7625 - accuracy: 0.3143 - val_loss: 5.2061 - val_accuracy: 0.6000\n",
      "Epoch 20/50\n",
      "70/70 [==============================] - 0s 5ms/sample - loss: 23.9521 - accuracy: 0.2286 - val_loss: 5.0559 - val_accuracy: 0.6000\n",
      "Epoch 21/50\n",
      "70/70 [==============================] - 0s 5ms/sample - loss: 16.1879 - accuracy: 0.3286 - val_loss: 4.8379 - val_accuracy: 0.5667\n",
      "Epoch 22/50\n",
      "70/70 [==============================] - 0s 5ms/sample - loss: 15.2995 - accuracy: 0.3000 - val_loss: 4.7751 - val_accuracy: 0.5333\n",
      "Epoch 23/50\n",
      "70/70 [==============================] - 0s 5ms/sample - loss: 19.4631 - accuracy: 0.3000 - val_loss: 4.6781 - val_accuracy: 0.6000\n",
      "Epoch 24/50\n",
      "70/70 [==============================] - 0s 6ms/sample - loss: 19.4401 - accuracy: 0.3143 - val_loss: 4.5940 - val_accuracy: 0.5333\n",
      "Epoch 25/50\n",
      "70/70 [==============================] - 0s 5ms/sample - loss: 18.2504 - accuracy: 0.3143 - val_loss: 4.7444 - val_accuracy: 0.5667\n",
      "Epoch 26/50\n",
      "70/70 [==============================] - 0s 4ms/sample - loss: 15.9337 - accuracy: 0.3714 - val_loss: 4.8683 - val_accuracy: 0.5667\n",
      "Epoch 27/50\n",
      "70/70 [==============================] - 0s 4ms/sample - loss: 14.0259 - accuracy: 0.3000 - val_loss: 5.0488 - val_accuracy: 0.5333\n",
      "Epoch 28/50\n",
      "70/70 [==============================] - 0s 6ms/sample - loss: 17.1310 - accuracy: 0.2857 - val_loss: 4.9449 - val_accuracy: 0.5333\n",
      "Epoch 29/50\n",
      "70/70 [==============================] - 0s 5ms/sample - loss: 14.3619 - accuracy: 0.3714 - val_loss: 4.6396 - val_accuracy: 0.5333\n",
      "Epoch 30/50\n",
      "70/70 [==============================] - 0s 4ms/sample - loss: 15.3938 - accuracy: 0.3857 - val_loss: 4.5024 - val_accuracy: 0.5333\n",
      "Epoch 31/50\n",
      "70/70 [==============================] - 0s 5ms/sample - loss: 15.8950 - accuracy: 0.3429 - val_loss: 4.3535 - val_accuracy: 0.5000\n",
      "Epoch 32/50\n",
      "70/70 [==============================] - 0s 4ms/sample - loss: 15.9132 - accuracy: 0.4000 - val_loss: 4.2922 - val_accuracy: 0.5333\n",
      "Epoch 33/50\n",
      "70/70 [==============================] - 0s 5ms/sample - loss: 9.9341 - accuracy: 0.4714 - val_loss: 4.4627 - val_accuracy: 0.5333\n",
      "Epoch 34/50\n",
      "70/70 [==============================] - 0s 5ms/sample - loss: 13.1170 - accuracy: 0.3286 - val_loss: 4.5665 - val_accuracy: 0.5000\n",
      "Epoch 35/50\n",
      "70/70 [==============================] - 0s 5ms/sample - loss: 11.8010 - accuracy: 0.3286 - val_loss: 4.7335 - val_accuracy: 0.5000\n",
      "Epoch 36/50\n",
      "70/70 [==============================] - 0s 6ms/sample - loss: 10.5212 - accuracy: 0.4571 - val_loss: 4.7394 - val_accuracy: 0.5333\n",
      "Epoch 37/50\n",
      "70/70 [==============================] - 0s 4ms/sample - loss: 9.4711 - accuracy: 0.4714 - val_loss: 4.5838 - val_accuracy: 0.5000\n",
      "Epoch 38/50\n",
      "70/70 [==============================] - 0s 5ms/sample - loss: 10.9664 - accuracy: 0.4571 - val_loss: 4.4164 - val_accuracy: 0.5667\n",
      "Epoch 39/50\n",
      "70/70 [==============================] - 0s 6ms/sample - loss: 10.6467 - accuracy: 0.5143 - val_loss: 4.2628 - val_accuracy: 0.5667\n",
      "Epoch 40/50\n",
      "70/70 [==============================] - 0s 5ms/sample - loss: 9.8797 - accuracy: 0.4000 - val_loss: 4.1431 - val_accuracy: 0.5333\n",
      "Epoch 41/50\n",
      "70/70 [==============================] - 1s 7ms/sample - loss: 10.6453 - accuracy: 0.4143 - val_loss: 3.9830 - val_accuracy: 0.6000\n",
      "Epoch 42/50\n",
      "70/70 [==============================] - 0s 6ms/sample - loss: 7.3364 - accuracy: 0.5429 - val_loss: 3.8924 - val_accuracy: 0.5667\n",
      "Epoch 43/50\n",
      "70/70 [==============================] - 0s 7ms/sample - loss: 10.2123 - accuracy: 0.4286 - val_loss: 3.6619 - val_accuracy: 0.5333\n",
      "Epoch 44/50\n",
      "70/70 [==============================] - 0s 7ms/sample - loss: 8.1087 - accuracy: 0.4286 - val_loss: 3.4952 - val_accuracy: 0.5000\n",
      "Epoch 45/50\n",
      "70/70 [==============================] - 0s 6ms/sample - loss: 10.4024 - accuracy: 0.5000 - val_loss: 3.4664 - val_accuracy: 0.5000\n",
      "Epoch 46/50\n",
      "70/70 [==============================] - 0s 7ms/sample - loss: 11.6550 - accuracy: 0.4571 - val_loss: 3.5374 - val_accuracy: 0.5000\n",
      "Epoch 47/50\n",
      "70/70 [==============================] - 1s 8ms/sample - loss: 9.4701 - accuracy: 0.4286 - val_loss: 3.4951 - val_accuracy: 0.5333\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 48/50\n",
      "70/70 [==============================] - 0s 6ms/sample - loss: 7.8861 - accuracy: 0.5429 - val_loss: 3.4460 - val_accuracy: 0.4667\n",
      "Epoch 49/50\n",
      "70/70 [==============================] - 0s 6ms/sample - loss: 8.0741 - accuracy: 0.5143 - val_loss: 3.4552 - val_accuracy: 0.5000\n",
      "Epoch 50/50\n",
      "70/70 [==============================] - 0s 5ms/sample - loss: 10.8888 - accuracy: 0.4286 - val_loss: 3.4091 - val_accuracy: 0.5333\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "if __name__ == \"__main__\":\n",
    "    inputs, targets = load_data(DATASET_PATH)\n",
    "    \n",
    "    #split into train & test\n",
    "    inputs_train, inputs_test, targets_train, targets_test = train_test_split(inputs,\n",
    "                                                                             targets,\n",
    "                                                                             test_size=0.3)\n",
    "    \n",
    "    #build network architecture\n",
    "    #a sequential model - 1 input...3 hidden ..1 output layer\n",
    "    model = keras.Sequential([\n",
    "        #input layer\n",
    "        keras.layers.Flatten(input_shape=(inputs.shape[1], inputs.shape[2])), #2D array of mfcc [intervals, mffc value for interval]\n",
    "        \n",
    "        #1st hidden layer\n",
    "        keras.layers.Dense(512, activation=\"relu\", kernel_regularizer=keras.regularizers.l2(0.001)), #regularizer-lambda\n",
    "        keras.layers.Dropout(0.3),\n",
    "        \n",
    "        #2nd hidden layer\n",
    "        keras.layers.Dense(256, activation=\"relu\",kernel_regularizer=keras.regularizers.l2(0.001)),\n",
    "        keras.layers.Dropout(0.3),\n",
    "        \n",
    "        #3rd hidden layer\n",
    "        keras.layers.Dense(64, activation=\"relu\",kernel_regularizer=keras.regularizers.l2(0.001)),\n",
    "        keras.layers.Dropout(0.3),\n",
    "        \n",
    "        #multiclass classification problem\n",
    "        #output layer\n",
    "        keras.layers.Dense(10, activation=\"softmax\") #enables us to normalize all 10 output neurons(genres) (sum=1)\n",
    "    ])\n",
    "    \n",
    "    \n",
    "    #compile network\n",
    "    optimizer = keras.optimizers.Adam(learning_rate=0.0001)\n",
    "    model.compile(optimizer = optimizer,\n",
    "                 loss=\"sparse_categorical_crossentropy\",\n",
    "                 metrics=[\"accuracy\"])\n",
    "    model.summary()\n",
    "    \n",
    "    \n",
    "    #train network\n",
    "    history = model.fit(inputs_train, targets_train, \n",
    "              validation_data=(inputs_test, targets_test),\n",
    "              epochs=50,\n",
    "              batch_size=32) \n",
    "#batch --- stochastic(quick/innacurate)...full batch (gradient on all --- Slow/Accurate/MemoryIntensive)...mini-batch(X samples)\n",
    "    \n",
    "    #plot error and accuracy over the epochs\n",
    "    plot_history(history)"
   ]
  }
 ],
 "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": "3.7.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}