Sound_Classifier_Neural_Network.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Sound_Classifier_Neural_Network"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "'2.2.0'"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import glob\n",
    "import os\n",
    "import librosa\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "%matplotlib inline\n",
    "\n",
    "plt.style.use('ggplot')\n",
    "\n",
    "import keras\n",
    "from keras import models\n",
    "from keras.layers import Dense, Dropout, Activation, Flatten\n",
    "from keras.layers import Convolution2D, MaxPooling2D\n",
    "from keras.utils import to_categorical\n",
    "keras.__version__"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'../data/UrbanSound8K/audio/train\\\\100652-3-0-1.wav'"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "spl = 0\n",
    "\n",
    "parent_dir = '../data/UrbanSound8K/audio/'\n",
    "if spl:\n",
    "    sub_dir = 'samples/'\n",
    "else:\n",
    "    sub_dir = 'train/'    \n",
    "file_name = '*.wav'\n",
    "files = glob.glob(os.path.join(parent_dir, sub_dir, file_name))\n",
    "test_files = glob.glob(os.path.join(parent_dir, 'fold10/', file_name))\n",
    "files[1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "def parse_audio(files):\n",
    "    features, labels = np.empty((0,52)), np.empty(0)\n",
    "    for file in files:\n",
    "        # extract label from filename\n",
    "        labels = np.append(labels, file.split('\\\\')[1].split('-')[1])\n",
    "        \n",
    "        # extract features from audio file\n",
    "        x, sr = librosa.load(file)\n",
    "        stft = np.abs(librosa.stft(x))\n",
    "        mfccs = np.mean(librosa.feature.mfcc(y=x, sr=sr, n_mfcc=40).T, axis=0)\n",
    "        chroma = np.mean(librosa.feature.chroma_stft(S=stft,sr=sr).T, axis=0)\n",
    "        feature = np.hstack((mfccs,chroma))\n",
    "#         print(chroma.shape)\n",
    "#         print(mfccs.shape)\n",
    "#         print(feature.shape)\n",
    "        features = np.vstack((features, feature))\n",
    "    return np.array(features), np.array(labels, dtype=np.int)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "features, labels = parse_audio(files)\n",
    "train_labels = to_categorical(labels)\n",
    "print(features.shape)\n",
    "print(labels.shape)\n",
    "\n",
    "test_features, test_labels = parse_audio(test_files)\n",
    "test_labels = to_categorical(test_labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "np.savetxt('train_features',features)\n",
    "np.savetxt('train_labels',train_labels)\n",
    "np.savetxt('test_features',test_features)\n",
    "np.savetxt('test_labels',test_labels)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Build and compile model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "# model = models.Sequential()\n",
    "# model.add(layers.Dense(256, activation='relu', input_shape=(52,)))\n",
    "# model.add(layers.Dense(50, activation='relu'))\n",
    "# # model.add(layers.Dense(512, activation='relu'))\n",
    "# model.add(layers.Dense(10, activation='softmax'))\n",
    "\n",
    "# model.compile(optimizer='Adagrad',\n",
    "#                 loss='categorical_crossentropy',\n",
    "#                 metrics=['accuracy'])\n",
    "\n",
    "model = models.Sequential()\n",
    "\n",
    "model.add(Dense(256,input_shape=(52,)))\n",
    "model.add(Activation('relu'))\n",
    "model.add(Dropout(0.5))\n",
    "\n",
    "model.add(Dense(256))\n",
    "model.add(Activation('relu'))\n",
    "model.add(Dropout(0.5))\n",
    "\n",
    "model.add(Dense(256))\n",
    "model.add(Activation('relu'))\n",
    "model.add(Dropout(0.5))\n",
    "\n",
    "model.add(Dense(256))\n",
    "model.add(Activation('relu'))\n",
    "model.add(Dropout(0.5))\n",
    "\n",
    "model.add(Dense(10))\n",
    "model.add(Activation('softmax'))\n",
    "\n",
    "model.compile(optimizer='Adam',\n",
    "                loss='categorical_crossentropy',\n",
    "                metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Train model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 2686 samples, validate on 837 samples\n",
      "Epoch 1/50\n",
      "2686/2686 [==============================] - 2s 775us/step - loss: 13.1498 - acc: 0.1322 - val_loss: 8.5091 - val_acc: 0.2593\n",
      "Epoch 2/50\n",
      "2686/2686 [==============================] - 1s 242us/step - loss: 12.8031 - acc: 0.1608 - val_loss: 11.7348 - val_acc: 0.1589\n",
      "Epoch 3/50\n",
      "2686/2686 [==============================] - 1s 242us/step - loss: 11.8480 - acc: 0.1951 - val_loss: 9.7277 - val_acc: 0.2091\n",
      "Epoch 4/50\n",
      "2686/2686 [==============================] - 1s 240us/step - loss: 9.6323 - acc: 0.2074 - val_loss: 2.9280 - val_acc: 0.2151\n",
      "Epoch 5/50\n",
      "2686/2686 [==============================] - 1s 294us/step - loss: 4.0753 - acc: 0.1966 - val_loss: 2.2166 - val_acc: 0.2521\n",
      "Epoch 6/50\n",
      "2686/2686 [==============================] - 1s 268us/step - loss: 2.4666 - acc: 0.2211 - val_loss: 2.1852 - val_acc: 0.1995\n",
      "Epoch 7/50\n",
      "2686/2686 [==============================] - 1s 278us/step - loss: 2.2166 - acc: 0.2405 - val_loss: 2.1598 - val_acc: 0.2330\n",
      "Epoch 8/50\n",
      "2686/2686 [==============================] - 1s 247us/step - loss: 2.1291 - acc: 0.2580 - val_loss: 2.1237 - val_acc: 0.2748\n",
      "Epoch 9/50\n",
      "2686/2686 [==============================] - 1s 302us/step - loss: 2.0336 - acc: 0.2856 - val_loss: 2.0454 - val_acc: 0.3178\n",
      "Epoch 10/50\n",
      "2686/2686 [==============================] - 1s 251us/step - loss: 2.0170 - acc: 0.2967 - val_loss: 2.0451 - val_acc: 0.2915\n",
      "Epoch 11/50\n",
      "2686/2686 [==============================] - 1s 281us/step - loss: 1.9170 - acc: 0.3362 - val_loss: 1.9840 - val_acc: 0.3297\n",
      "Epoch 12/50\n",
      "2686/2686 [==============================] - 1s 298us/step - loss: 1.8561 - acc: 0.3574 - val_loss: 1.9121 - val_acc: 0.3501\n",
      "Epoch 13/50\n",
      "2686/2686 [==============================] - 1s 271us/step - loss: 1.7711 - acc: 0.3909 - val_loss: 1.8688 - val_acc: 0.3584\n",
      "Epoch 14/50\n",
      "2686/2686 [==============================] - 1s 286us/step - loss: 1.7446 - acc: 0.3943 - val_loss: 1.8162 - val_acc: 0.4098\n",
      "Epoch 15/50\n",
      "2686/2686 [==============================] - 1s 390us/step - loss: 1.6430 - acc: 0.4267 - val_loss: 1.7411 - val_acc: 0.4731\n",
      "Epoch 16/50\n",
      "2686/2686 [==============================] - 1s 413us/step - loss: 1.6218 - acc: 0.4401 - val_loss: 1.7364 - val_acc: 0.4421\n",
      "Epoch 17/50\n",
      "2686/2686 [==============================] - 1s 388us/step - loss: 1.5416 - acc: 0.4751 - val_loss: 1.7072 - val_acc: 0.4409\n",
      "Epoch 18/50\n",
      "2686/2686 [==============================] - 1s 407us/step - loss: 1.4907 - acc: 0.4829 - val_loss: 1.6668 - val_acc: 0.4217\n",
      "Epoch 19/50\n",
      "2686/2686 [==============================] - 1s 428us/step - loss: 1.4308 - acc: 0.5112 - val_loss: 1.6261 - val_acc: 0.4492\n",
      "Epoch 20/50\n",
      "2686/2686 [==============================] - 1s 391us/step - loss: 1.3739 - acc: 0.5261 - val_loss: 1.6281 - val_acc: 0.4922\n",
      "Epoch 21/50\n",
      "2686/2686 [==============================] - 1s 378us/step - loss: 1.3403 - acc: 0.5514 - val_loss: 1.6537 - val_acc: 0.4444\n",
      "Epoch 22/50\n",
      "2686/2686 [==============================] - 1s 421us/step - loss: 1.2957 - acc: 0.5506 - val_loss: 1.6364 - val_acc: 0.4516\n",
      "Epoch 23/50\n",
      "2686/2686 [==============================] - 1s 434us/step - loss: 1.2605 - acc: 0.5771 - val_loss: 1.6125 - val_acc: 0.5006\n",
      "Epoch 24/50\n",
      "2686/2686 [==============================] - 1s 473us/step - loss: 1.2356 - acc: 0.5663 - val_loss: 1.6059 - val_acc: 0.4839\n",
      "Epoch 25/50\n",
      "2686/2686 [==============================] - 1s 444us/step - loss: 1.1729 - acc: 0.5916 - val_loss: 1.6217 - val_acc: 0.4875\n",
      "Epoch 26/50\n",
      "2686/2686 [==============================] - 1s 463us/step - loss: 1.1612 - acc: 0.5953 - val_loss: 1.6144 - val_acc: 0.4456\n",
      "Epoch 27/50\n",
      "2686/2686 [==============================] - 1s 441us/step - loss: 1.1055 - acc: 0.6206 - val_loss: 1.6383 - val_acc: 0.4373\n",
      "Epoch 28/50\n",
      "2686/2686 [==============================] - 1s 414us/step - loss: 1.0957 - acc: 0.6210 - val_loss: 1.6738 - val_acc: 0.4851\n",
      "Epoch 29/50\n",
      "2686/2686 [==============================] - 1s 376us/step - loss: 1.0592 - acc: 0.6296 - val_loss: 1.6904 - val_acc: 0.4659\n",
      "Epoch 30/50\n",
      "2686/2686 [==============================] - 1s 382us/step - loss: 1.0491 - acc: 0.6512 - val_loss: 1.6667 - val_acc: 0.4970\n",
      "Epoch 31/50\n",
      "2686/2686 [==============================] - 1s 379us/step - loss: 1.0102 - acc: 0.6590 - val_loss: 1.6715 - val_acc: 0.4719\n",
      "Epoch 32/50\n",
      "2686/2686 [==============================] - 1s 398us/step - loss: 0.9938 - acc: 0.6660 - val_loss: 1.7188 - val_acc: 0.4910\n",
      "Epoch 33/50\n",
      "2686/2686 [==============================] - 1s 384us/step - loss: 0.9480 - acc: 0.6787 - val_loss: 1.6877 - val_acc: 0.4456\n",
      "Epoch 34/50\n",
      "2686/2686 [==============================] - 1s 408us/step - loss: 0.9410 - acc: 0.6727 - val_loss: 1.6739 - val_acc: 0.4516\n",
      "Epoch 35/50\n",
      "2686/2686 [==============================] - 1s 392us/step - loss: 0.9400 - acc: 0.6821 - val_loss: 1.6848 - val_acc: 0.4970\n",
      "Epoch 36/50\n",
      "2686/2686 [==============================] - 1s 405us/step - loss: 0.9268 - acc: 0.6917 - val_loss: 1.6262 - val_acc: 0.5114\n",
      "Epoch 37/50\n",
      "2686/2686 [==============================] - 1s 387us/step - loss: 0.8449 - acc: 0.7159 - val_loss: 1.7366 - val_acc: 0.5257\n",
      "Epoch 38/50\n",
      "2686/2686 [==============================] - 1s 382us/step - loss: 0.8960 - acc: 0.6999 - val_loss: 1.6705 - val_acc: 0.4815\n",
      "Epoch 39/50\n",
      "2686/2686 [==============================] - 1s 386us/step - loss: 0.8650 - acc: 0.7137 - val_loss: 1.6759 - val_acc: 0.5006\n",
      "Epoch 40/50\n",
      "2686/2686 [==============================] - 1s 379us/step - loss: 0.8479 - acc: 0.7163 - val_loss: 1.7041 - val_acc: 0.5006\n",
      "Epoch 41/50\n",
      "2686/2686 [==============================] - 1s 392us/step - loss: 0.8406 - acc: 0.7241 - val_loss: 1.6608 - val_acc: 0.4946\n",
      "Epoch 42/50\n",
      "2686/2686 [==============================] - 1s 396us/step - loss: 0.8141 - acc: 0.7297 - val_loss: 1.6669 - val_acc: 0.4875\n",
      "Epoch 43/50\n",
      "2686/2686 [==============================] - 1s 406us/step - loss: 0.7533 - acc: 0.7375 - val_loss: 1.7022 - val_acc: 0.4958\n",
      "Epoch 44/50\n",
      "2686/2686 [==============================] - 1s 429us/step - loss: 0.7606 - acc: 0.7427 - val_loss: 1.6468 - val_acc: 0.4910\n",
      "Epoch 45/50\n",
      "2686/2686 [==============================] - 1s 416us/step - loss: 0.7576 - acc: 0.7319 - val_loss: 1.7242 - val_acc: 0.5125\n",
      "Epoch 46/50\n",
      "2686/2686 [==============================] - 1s 411us/step - loss: 0.7575 - acc: 0.7502 - val_loss: 1.6251 - val_acc: 0.5209\n",
      "Epoch 47/50\n",
      "2686/2686 [==============================] - 1s 380us/step - loss: 0.7460 - acc: 0.7509 - val_loss: 1.6690 - val_acc: 0.5018\n",
      "Epoch 48/50\n",
      "2686/2686 [==============================] - 1s 384us/step - loss: 0.7294 - acc: 0.7494 - val_loss: 1.6261 - val_acc: 0.5137\n",
      "Epoch 49/50\n",
      "2686/2686 [==============================] - 1s 372us/step - loss: 0.7060 - acc: 0.7602 - val_loss: 1.6200 - val_acc: 0.5281\n",
      "Epoch 50/50\n",
      "2686/2686 [==============================] - 1s 388us/step - loss: 0.6886 - acc: 0.7535 - val_loss: 1.6549 - val_acc: 0.5257\n"
     ]
    }
   ],
   "source": [
    "# from keras.callbacks import TensorBoard\n",
    "# from keras.utils import plot_model\n",
    "\n",
    "# tb = TensorBoard(log_dir='./logs',  # log 目录\n",
    "#                  histogram_freq=1,  # 按照何等频率（epoch）来计算直方图，0为不计算\n",
    "#                  batch_size=32,     # 用多大量的数据计算直方图\n",
    "#                  write_graph=True,  # 是否存储网络结构图\n",
    "#                  write_grads=False, # 是否可视化梯度直方图\n",
    "#                  write_images=False,# 是否可视化参数\n",
    "#                  embeddings_freq=0, \n",
    "#                  embeddings_layer_names=None, \n",
    "#                  embeddings_metadata=None)\n",
    "# callbacks = [tb]\n",
    "# history = model.fit(features, train_labels, epochs=50, batch_size=32, validation_data=(test_features, test_labels),callbacks=callbacks)\n",
    "\n",
    "history = model.fit(features, train_labels, epochs=50, batch_size=32, validation_data=(test_features, test_labels))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "dict_keys(['val_loss', 'val_acc', 'loss', 'acc'])\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x360 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "print(history.history.keys())\n",
    "\n",
    "fig = plt.figure(figsize=(20,5))\n",
    "plt.subplot(121)\n",
    "plt.plot(history.history['acc'])\n",
    "plt.plot(history.history['val_acc'])\n",
    "plt.title('model accuracy')\n",
    "plt.ylabel('accuracy')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'test'], loc='upper left')\n",
    "plt.subplot(122)\n",
    "plt.plot(history.history['loss'])\n",
    "plt.plot(history.history['val_loss'])\n",
    "plt.title('model loss')\n",
    "plt.ylabel('loss')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'test'], loc='lower left')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Test model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "837/837 [==============================] - 0s 306us/step\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.5256869772998806"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_loss, test_acc = model.evaluate(test_features, test_labels)\n",
    "test_acc"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python (tensorflow)",
   "language": "python",
   "name": "tensorflow"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}