Removed long code error

training/.ipynb_checkpoints/standard_cnn_model-checkpoint.ipynb

@@ -0,0 +1,286 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import pandas as pd\n",
+    "import numpy as np\n",
+    "import os\n",
+    "import librosa\n",
+    "import librosa.display\n",
+    "import matplotlib.pyplot as plt\n",
+    "from sklearn.preprocessing import LabelEncoder\n",
+    "from keras.utils import to_categorical\n",
+    "import h5py\n",
+    "import math"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import struct\n",
+    "\n",
+    "class WavFileHelper():\n",
+    "    \n",
+    "    def read_file_properties(self, filename):\n",
+    "\n",
+    "        wave_file = open(filename,\"rb\")\n",
+    "        \n",
+    "        riff = wave_file.read(12)\n",
+    "        fmt = wave_file.read(36)\n",
+    "        \n",
+    "        num_channels_string = fmt[10:12]\n",
+    "        num_channels = struct.unpack('<H', num_channels_string)[0]\n",
+    "\n",
+    "        sample_rate_string = fmt[12:16]\n",
+    "        sample_rate = struct.unpack(\"<I\",sample_rate_string)[0]\n",
+    "        \n",
+    "        bit_depth_string = fmt[22:24]\n",
+    "        bit_depth = struct.unpack(\"<H\",bit_depth_string)[0]\n",
+    "\n",
+    "        return (num_channels, sample_rate, bit_depth)\n",
+    "\n",
+    "wavfilehelper = WavFileHelper()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def extract_features(file_name):\n",
+    "   \n",
+    "    try:\n",
+    "        audio_y, sample_rate = librosa.load(file_name, res_type='kaiser_best') \n",
+    "        D = (librosa.stft(audio_y, center=False, n_fft=1024))\n",
+    "        #print(D)\n",
+    "        real = np.real(D)\n",
+    "        imag = np.imag(D)\n",
+    "        D_true = np.stack([real, imag], -1)\n",
+    "        \n",
+    "        #print(D_true.shape)\n",
+    "        if(D_true.shape != (513, 341, 2)):\n",
+    "            #print(\"ALERTTT\")\n",
+    "            #print(D_true)\n",
+    "            num_of_times_to_repeat = math.ceil(341 / D_true.shape[1])\n",
+    "            D_true = np.repeat(D_true, num_of_times_to_repeat, axis=1)\n",
+    "            D_true = D_true[:, 0:341, :]\n",
+    "            #print(D_true.shape)\n",
+    "            #print(D_true)\n",
+    "        \n",
+    "        #result = np.zeros(b.shape)\n",
+    "\n",
+    "        \n",
+    "        #print(D_true.shape)\n",
+    "        \n",
+    "        #librosa.display.specshow(librosa.amplitude_to_db(D,\n",
+    "        #                                         ref=np.max),\n",
+    "        #                 y_axis='log', x_axis='time')\n",
+    "        #plt.title('Power spectrogram')\n",
+    "        #plt.colorbar(format='%+2.0f dB')\n",
+    "        #plt.tight_layout()\n",
+    "        #plt.show()\n",
+    "        \n",
+    "        \n",
+    "    except Exception as e:\n",
+    "        print(\"Error encountered while parsing file: \", file)\n",
+    "        return None \n",
+    "     \n",
+    "    return (D_true)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "metadata = pd.read_csv('./UrbanSound8k/metadata/UrbanSound8K.csv')\n",
+    "\n",
+    "audiodata = []\n",
+    "for index, row in metadata.iterrows():\n",
+    "    \n",
+    "    file_name = os.path.join(os.path.abspath('UrbanSound8k/audio/'),'fold'+str(row[\"fold\"])+'/',str(row[\"slice_file_name\"]))\n",
+    "    data = wavfilehelper.read_file_properties(file_name)\n",
+    "    audiodata.append(data)\n",
+    "\n",
+    "# Convert into a Panda dataframe\n",
+    "audiodf = pd.DataFrame(audiodata, columns=['num_channels','sample_rate','bit_depth'])"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "features = np.load(\"features.npy\", allow_pickle=True)\n",
+    "labels   = np.load(\"labels.npy\", allow_pickle=True)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "metadata = pd.read_csv('./UrbanSound8k/metadata/UrbanSound8K.csv')\n",
+    "features = []\n",
+    "labels = []\n",
+    "i = 0\n",
+    "\n",
+    "# Iterate through each sound file and extract the features \n",
+    "for index, row in metadata.iterrows():\n",
+    "    \n",
+    "    i = i +1\n",
+    "    file_name = os.path.join(os.path.abspath('UrbanSound8k/audio/'),'fold'+str(row[\"fold\"])+'/',str(row[\"slice_file_name\"]))\n",
+    "    \n",
+    "    class_label = row[\"class\"]\n",
+    "    print(f'{i}: Processing {file_name}')\n",
+    "    data = extract_features(file_name)\n",
+    "    \n",
+    "    features.append(data)\n",
+    "    labels.append(class_label)\n",
+    "\n",
+    "# Convert into a Panda dataframe \n",
+    "#featuresdf = pd.DataFrame(features, columns=['feature','class_label'])\n",
+    "np.save('features.npy', features)\n",
+    "np.save('labels.npy',   labels)\n",
+    "\n",
+    "\n",
+    "print('Finished feature extraction from ', len(features), ' files')\n",
+    "\n",
+    "X = np.asarray(features)\n",
+    "y = np.asarray(labels)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "X = (features)\n",
+    "y = (labels)\n",
+    "\n",
+    "# Encode the classification labels\n",
+    "le = LabelEncoder()\n",
+    "yy = to_categorical(le.fit_transform(y)) \n",
+    "\n",
+    "# split the dataset \n",
+    "from sklearn.model_selection import train_test_split \n",
+    "\n",
+    "x_train, x_test, y_train, y_test = train_test_split(X, yy, test_size=0.2, random_state = 42)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from keras.models import Sequential\n",
+    "from keras.layers import Dense, Dropout, Activation, Flatten\n",
+    "from keras.layers import Convolution2D, Conv2D, MaxPooling2D, GlobalAveragePooling2D\n",
+    "from keras.optimizers import Adam\n",
+    "from keras.utils import np_utils\n",
+    "from sklearn import metrics \n",
+    "\n",
+    "\n",
+    "num_labels = yy.shape[1]\n",
+    "\n",
+    "# Construct model \n",
+    "model = Sequential()\n",
+    "model.add(Conv2D(filters=16, kernel_size=2, input_shape=(513, 341, 2), activation='relu'))\n",
+    "model.add(MaxPooling2D(pool_size=2))\n",
+    "model.add(Dropout(0.2))\n",
+    "\n",
+    "model.add(Conv2D(filters=32, kernel_size=2, activation='relu'))\n",
+    "model.add(MaxPooling2D(pool_size=2))\n",
+    "model.add(Dropout(0.2))\n",
+    "\n",
+    "model.add(Conv2D(filters=64, kernel_size=2, activation='relu'))\n",
+    "model.add(MaxPooling2D(pool_size=2))\n",
+    "model.add(Dropout(0.2))\n",
+    "\n",
+    "model.add(GlobalAveragePooling2D())\n",
+    "\n",
+    "model.add(Dense(num_labels, activation='softmax'))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "model.compile(loss='categorical_crossentropy', metrics=['accuracy'], optimizer='adam')\n",
+    "model.summary()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "score = model.evaluate(x_test, y_test, verbose=1)\n",
+    "accuracy = 100*score[1]\n",
+    "\n",
+    "print(\"Pre-training accuracy: %.4f%%\" % accuracy) "
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from keras.callbacks import ModelCheckpoint, TensorBoard\n",
+    "from datetime import datetime \n",
+    "\n",
+    "num_epochs = 72\n",
+    "num_batch_size = 256\n",
+    "\n",
+    "checkpointer = ModelCheckpoint(filepath='saved_models/weights.best.basic_cnn.hdf5', \n",
+    "                               verbose=1, save_best_only=True)\n",
+    "start = datetime.now()\n",
+    "\n",
+    "tbCallBack = TensorBoard(log_dir='./tensorboard_graphs', histogram_freq=0, write_graph=True, write_images=True)\n",
+    "model.fit(x_train, y_train, batch_size=num_batch_size, epochs=num_epochs, validation_data=(x_test, y_test), callbacks=[checkpointer, tbCallBack], verbose=1)\n",
+    "\n",
+    "\n",
+    "duration = datetime.now() - start\n",
+    "print(\"Training completed in time: \", duration)"
+   ]
+  }
+ ],
+ "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": "3.8.6"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}

training/standard_cnn_model.ipynb

@@ -278,7 +278,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.7.7"
+   "version": "3.8.6"
   }
  },
  "nbformat": 4,