RAVDESS-CNN-MFCC.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# speech emotion classification\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import librosa\n",
    "import librosa.display\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### File naming convention\n",
    "Each of the 1440 files has a unique filename. The filename consists of a 7-part numerical identifier (e.g., 03-01-06-01-02-01-12.wav). These identifiers define the stimulus characteristics:\n",
    "<br>\n",
    "##### Filename identifiers\n",
    "- Modality (01 = full-AV, 02 = video-only, 03 = audio-only).\n",
    "<br>\n",
    "- Vocal channel (01 = speech, 02 = song).\n",
    "<br>\n",
    "- Emotion (01 = neutral, 02 = calm, 03 = happy, 04 = sad, 05 = angry, 06 = fearful, 07 = disgust, 08 = surprised).\n",
    "<br>\n",
    "- Emotional intensity (01 = normal, 02 = strong). NOTE: There is no strong intensity for the 'neutral' emotion.\n",
    "<br>\n",
    "- Statement (01 = \"Kids are talking by the door\", 02 = \"Dogs are sitting by the door\").\n",
    "<br>\n",
    "- Repetition (01 = 1st repetition, 02 = 2nd repetition).\n",
    "<br>\n",
    "- Actor (01 to 24. Odd numbered actors are male, even numbered actors are female).\n",
    "<br>\n",
    "\n",
    "##### Filename example: 03-01-06-01-02-01-12.wav\n",
    "- Audio-only (03)\n",
    "<br>\n",
    "- Speech (01)\n",
    "<br>\n",
    "- Fearful (06)\n",
    "<br>\n",
    "- Normal intensity (01)\n",
    "<br>\n",
    "- Statement \"dogs\" (02)\n",
    "<br>\n",
    "- 1st Repetition (01)\n",
    "<br>\n",
    "- 12th Actor (12) Female, as the actor ID number is even."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Emotion labels\n",
    "01 = neutral\n",
    "<br>\n",
    "02 = calm\n",
    "<br>\n",
    "03 = happy\n",
    "<br>\n",
    "04 = sad \n",
    "<br>\n",
    "05 = angry \n",
    "<br>\n",
    "06 = fearful\n",
    "<br>\n",
    "07 = disgust\n",
    "<br>\n",
    "08 = surprised"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data usage\n",
    "in this notebook we will use the following elements:\n",
    "- Modality (03 = audio-only)\n",
    "<br>\n",
    "- Vocal channel (01 = speech)\n",
    "<br>\n",
    "- Emotion (01 = neutral, 02 = calm, 03 = happy, 04 = sad, 05 = angry, 06 = fearful, 07 = disgust, 08 = surprised)\n",
    "<br>\n",
    "- Emotional intensity (01 = normal, 02 = strong)\n",
    "<br>\n",
    "- Statement (01 = \"Kids are talking by the door\")\n",
    "<br>\n",
    "- Repetition (01 = 1st repetition)\n",
    "<br>\n",
    "- Actor (01 to 24. Odd numbered actors are male, even numbered actors are female)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## step 1: Open RAVDESS dataset & explore structure of wave samples\n",
    "##### requirements:\n",
    "- Download Ravdess-RawData\n",
    "- Replace my_path with your own path"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "my_path = 'C:/Users/voice/RAVDESS-RawData'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "this is the len of our data:  1440\n",
      "this is an example of a wave file: 03-01-01-01-02-01-03.wav\n",
      "this index position represents the emotion label: 01\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "my_data = os.listdir(my_path)\n",
    "print(\"this is the len of our data: \", len(my_data))\n",
    "print(\"this is an example of a wave file:\", my_data[50])\n",
    "print(\"this index position represents the emotion label:\", my_data[50][6:8])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## step 2: Load audio file and visualize its waveform (using librosa)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "data, sampling_rate = librosa.load(my_path+'/03-01-01-01-02-01-03.wav')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PolyCollection at 0x28511ec0668>"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(15, 5))\n",
    "librosa.display.waveplot(data, sr=sampling_rate)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Step 3: Convert the audio waveform"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Mel-frequency cepstral coefficients (MFCC) extraction\n",
    "function returns:\n",
    "- features\n",
    "<br>\n",
    "- labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "from python_speech_features import mfcc\n",
    "import scipy.io.wavfile\n",
    "import numpy as np\n",
    "\n",
    "\n",
    "emotions_num = [\"Neutral-01\",\"Calm-02\",\"Happy-03\",\"Sad-04\",\"Angry-05\",\"Fearful-06\",\"Disgust-07\",\"Surprised-08\"]\n",
    "actors = [\"01\",\"02\",\"03\",\"04\",\"05\",\"06\",\"07\",\"08\",\"09\",\"10\",\"11\",\"12\",\n",
    "         \"13\",\"14\",\"15\",\"16\",\"17\",\"18\",\"19\",\"20\",\"21\",\"22\",\"23\",\"24\"]\n",
    "\n",
    "\n",
    "def mfcc_extraction(path, modal_vocal, emotions_num, actors):\n",
    "    labels = []\n",
    "    mfcc_features = []\n",
    "    \n",
    "    #with repetition number: 1 and statement number 1 and emotional intensity number #1\n",
    "    for index1,emotion_num in enumerate(emotions_num):\n",
    "        for index,actor in enumerate(actors):\n",
    "            sig, rate = librosa.load(path+modal_vocal+emotion_num[-2:]+\"-01-01-01-\"+actor+\".wav\", sr=16000)\n",
    "            mfcc_feat = mfcc(sig,rate)\n",
    "            mfcc_features.append(mfcc_feat)\n",
    "            \n",
    "            labels.append(emotion_num[:-3])\n",
    "    \n",
    "    #with repetition number: 1 and statement number 2 and emotional intensity number 1\n",
    "    for index1,emotion_num in enumerate(emotions_num):\n",
    "        for index,actor in enumerate(actors):\n",
    "            sig, rate = librosa.load(path+modal_vocal+emotion_num[-2:]+\"-01-02-01-\"+actor+\".wav\", sr=16000)\n",
    "            mfcc_feat = mfcc(sig,rate)\n",
    "            mfcc_features.append(mfcc_feat)\n",
    "            \n",
    "            labels.append(emotion_num[:-3])\n",
    "            \n",
    "    #with repetition number: 2 and statement number 1 and emotional intensity number #1\n",
    "    for index1,emotion_num in enumerate(emotions_num):\n",
    "        for index,actor in enumerate(actors):\n",
    "            sig, rate = librosa.load(path+modal_vocal+emotion_num[-2:]+\"-01-01-02-\"+actor+\".wav\", sr=16000)\n",
    "            mfcc_feat = mfcc(sig,rate)\n",
    "            mfcc_features.append(mfcc_feat)\n",
    "            \n",
    "            labels.append(emotion_num[:-3])\n",
    "            \n",
    "    #with repetition number: 2 and statement number 2 and emotional intensity number 1\n",
    "    for index1,emotion_num in enumerate(emotions_num):\n",
    "        for index,actor in enumerate(actors):\n",
    "            sig, rate = librosa.load(path+modal_vocal+emotion_num[-2:]+\"-01-02-02-\"+actor+\".wav\", sr=16000)\n",
    "            mfcc_feat = mfcc(sig,rate)\n",
    "            mfcc_features.append(mfcc_feat)\n",
    "            \n",
    "            labels.append(emotion_num[:-3])\n",
    "    \n",
    "    #with repetition number: 1 and statement number 1 and emotional intensity number #2\n",
    "    for index1,emotion_num in enumerate(emotions_num):\n",
    "        for index,actor in enumerate(actors):\n",
    "            if emotion_num != \"Neutral-01\":\n",
    "                sig, rate = librosa.load(path+modal_vocal+emotion_num[-2:]+\"-02-01-01-\"+actor+\".wav\", sr=16000)\n",
    "                mfcc_feat = mfcc(sig,rate)\n",
    "                mfcc_features.append(mfcc_feat)\n",
    "                \n",
    "                labels.append(emotion_num[:-3])\n",
    "    \n",
    "    #with repetition number: 1 and statement number 2 and emotional intensity number 2\n",
    "    for index1,emotion_num in enumerate(emotions_num):\n",
    "        for index,actor in enumerate(actors):\n",
    "            if emotion_num != \"Neutral-01\":\n",
    "                sig, rate = librosa.load(path+modal_vocal+emotion_num[-2:]+\"-02-02-01-\"+actor+\".wav\", sr=16000)\n",
    "                mfcc_feat = mfcc(sig,rate)\n",
    "                mfcc_features.append(mfcc_feat)\n",
    "                \n",
    "                labels.append(emotion_num[:-3])\n",
    "            \n",
    "    #with repetition number: 2 and statement number 1 and emotional intensity number #2\n",
    "    for index1,emotion_num in enumerate(emotions_num):\n",
    "        for index,actor in enumerate(actors):\n",
    "            if emotion_num != \"Neutral-01\":\n",
    "                sig, rate = librosa.load(path+modal_vocal+emotion_num[-2:]+\"-02-01-02-\"+actor+\".wav\", sr=16000)\n",
    "                mfcc_feat = mfcc(sig,rate)\n",
    "                mfcc_features.append(mfcc_feat)\n",
    "                \n",
    "                labels.append(emotion_num[:-3])\n",
    "            \n",
    "    #with repetition number: 2 and statement number 2 and emotional intensity number 2\n",
    "    for index1,emotion_num in enumerate(emotions_num):\n",
    "        for index,actor in enumerate(actors):\n",
    "            if emotion_num != \"Neutral-01\":\n",
    "                sig, rate = librosa.load(path+modal_vocal+emotion_num[-2:]+\"-02-02-02-\"+actor+\".wav\", sr=16000)\n",
    "                mfcc_feat = mfcc(sig,rate)\n",
    "                mfcc_features.append(mfcc_feat)\n",
    "                \n",
    "                labels.append(emotion_num[:-3])      \n",
    "\n",
    "    return mfcc_features, labels\n",
    "\n",
    "\n",
    "mfcc_features_and_labels = mfcc_extraction(my_path,'/03-01-', emotions_num, actors)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### In our function we are extracting 2(repetition)2(statement)2(emotional)8(emotions)24(actors)\n",
    "#### 1440 examples"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "len of examples: 1440\n",
      "len of labels: 1440\n",
      "\n",
      "feature dimension: 2\n",
      "feature shape: (363, 13)\n",
      "\n",
      "[[-21.83060237 -32.0043785    1.63623961 ...  15.17986358 -11.67160094\n",
      "   -2.84455921]\n",
      " [-36.04365339   0.           0.         ...   0.           0.\n",
      "    0.        ]\n",
      " [-23.18079535 -35.02521792 -10.93098371 ...  -5.13416197  -3.13953157\n",
      "   -4.02089669]\n",
      " ...\n",
      " [-20.12058616 -29.7380244   -0.1719999  ...   4.60930548   3.42111213\n",
      "   -1.32853321]\n",
      " [-22.21191914 -30.8699543   -7.99719638 ...  -0.61797758   2.82549932\n",
      "  -10.76478004]\n",
      " [-36.04365339   0.           0.         ...   0.           0.\n",
      "    0.        ]] Neutral\n"
     ]
    }
   ],
   "source": [
    "# exract features and labels from \n",
    "mfcc_features, mfcc_labels = mfcc_features_and_labels\n",
    "print(\"len of examples:\", len(mfcc_features))\n",
    "print(\"len of labels:\", len(mfcc_labels))\n",
    "print()\n",
    "print(\"feature dimension:\", mfcc_features[0].ndim)\n",
    "print(\"feature shape:\", mfcc_features[1].shape)\n",
    "print()\n",
    "mfcc_labels = np.array(mfcc_labels)\n",
    "print(mfcc_features[0],mfcc_labels[0])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# zero pad function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "def zero_pad(features):\n",
    "    \n",
    "    new_spoken_train = []\n",
    "    max_shape = 0\n",
    "    for x in features:\n",
    "        all_shapes = x.shape[0]\n",
    "        if all_shapes > max_shape:\n",
    "            max_shape = all_shapes\n",
    "\n",
    "    for x in features:\n",
    "        difference = max_shape-x.shape[0]\n",
    "        zero_padded = np.pad(x,((0,difference),(0,0)), \"constant\")\n",
    "        new_spoken_train.append(zero_padded)\n",
    "\n",
    "    new_spoken_train = np.array(new_spoken_train)\n",
    "    \n",
    "    return new_spoken_train\n",
    "\n",
    "mfcc_features = zero_pad(mfcc_features)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1440, 526, 13)\n",
      "6838\n"
     ]
    }
   ],
   "source": [
    "print(mfcc_features.shape)\n",
    "print(526*13)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Flatten MFCC features "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1440, 6838)\n"
     ]
    }
   ],
   "source": [
    "flatten_mfcc_features = np.reshape(mfcc_features,(mfcc_features.shape[0],6838))\n",
    "print(flatten_mfcc_features.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Step 4: standardize"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import StandardScaler\n",
    "\n",
    "data = flatten_mfcc_features\n",
    "scaler = StandardScaler()\n",
    "mfcc_features = scaler.fit_transform(data)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Step 5 oversamle neutral in order to match our dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'Neutral': 96, 'Calm': 192, 'Happy': 192, 'Sad': 192, 'Angry': 192, 'Fearful': 192, 'Disgust': 192, 'Surprised': 192}\n"
     ]
    }
   ],
   "source": [
    "dictionary = {}\n",
    "for i in mfcc_labels:\n",
    "    dictionary[i] = dictionary.get(i,0) + 1\n",
    "\n",
    "print(dictionary)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'Neutral': 192, 'Calm': 192, 'Happy': 192, 'Sad': 192, 'Angry': 192, 'Fearful': 192, 'Disgust': 192, 'Surprised': 192}\n"
     ]
    }
   ],
   "source": [
    "from imblearn.over_sampling import RandomOverSampler\n",
    "ros = RandomOverSampler(random_state=42)\n",
    "new_mfcc_features,  mfcc_labels = ros.fit_resample(mfcc_features,  mfcc_labels)\n",
    "\n",
    "dictionary_2 = {}\n",
    "for i in mfcc_labels:\n",
    "    dictionary_2[i] = dictionary_2.get(i,0) + 1\n",
    "\n",
    "print(dictionary_2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Reshape back mfcc features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "new_mfcc_features = np.reshape(new_mfcc_features,(new_mfcc_features.shape[0],526, 13))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## checkpoint"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1536, 526, 13)"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "new_mfcc_features.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# step 6: Split data in Train, Val & Test\n",
    "- features = stft_features\n",
    "- labels = categorical_stft_labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "X_train, X_test, y_train, y_test = train_test_split(new_mfcc_features, mfcc_labels, test_size=0.30, \n",
    "                                                    random_state=42, stratify = mfcc_labels, shuffle= True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## one-hot encode labels (X_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1075, 8)\n"
     ]
    }
   ],
   "source": [
    "from numpy import array\n",
    "from numpy import argmax\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "from sklearn.preprocessing import OneHotEncoder\n",
    "\n",
    "# define example\n",
    "y_train = y_train\n",
    "y_train_values = array(y_train)\n",
    "# integer encode\n",
    "y_train_label_encoder = LabelEncoder()\n",
    "y_train_integer_encoded = y_train_label_encoder.fit_transform(y_train_values)\n",
    "# binary encode\n",
    "y_train_onehot_encoder = OneHotEncoder(sparse=False)\n",
    "y_train_integer_encoded = y_train_integer_encoded.reshape(len(y_train_integer_encoded), 1)\n",
    "y_train_onehot_encoded = y_train_onehot_encoder.fit_transform(y_train_integer_encoded)\n",
    "print(y_train_onehot_encoded.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## one-hot encode labels (X_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(461, 8)\n"
     ]
    }
   ],
   "source": [
    "# define example\n",
    "y_test = y_test\n",
    "y_test_values = array(y_test)\n",
    "# integer encode\n",
    "y_test_label_encoder = LabelEncoder()\n",
    "y_test_integer_encoded = y_test_label_encoder.fit_transform(y_test_values)\n",
    "# binary encode\n",
    "y_test_onehot_encoder = OneHotEncoder(sparse=False)\n",
    "y_test_integer_encoded = y_test_integer_encoded.reshape(len(y_test_integer_encoded), 1)\n",
    "y_test_onehot_encoded = y_test_onehot_encoder.fit_transform(y_test_integer_encoded)\n",
    "print(y_test_onehot_encoded.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## checkpoint"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1075, 526, 13)\n",
      "(1075, 8)\n",
      "(461, 526, 13)\n",
      "(461, 8)\n"
     ]
    }
   ],
   "source": [
    "print(X_train.shape)\n",
    "print(y_train_onehot_encoded.shape)\n",
    "print(X_test.shape)\n",
    "print(y_test_onehot_encoded.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Convolutional Neural network"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "import keras\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense, Embedding\n",
    "from keras.layers import Input, Flatten, Dropout, Activation\n",
    "from keras.layers import Conv1D, MaxPooling1D\n",
    "from keras.models import Model\n",
    "from keras.callbacks import ModelCheckpoint\n",
    "\n",
    "len_of_labels = len(y_train_onehot_encoded[0])\n",
    "model = Sequential()\n",
    "\n",
    "model.add(Conv1D(128, 4, padding='same', input_shape=(X_train.shape[1],X_train.shape[2])))\n",
    "model.add(Activation('relu'))\n",
    "model.add(MaxPooling1D(pool_size=(2)))\n",
    "model.add(Conv1D(128, 4, padding='same',))\n",
    "model.add(Activation('relu'))\n",
    "model.add(Dropout(0.4))\n",
    "model.add(MaxPooling1D(pool_size=(2)))\n",
    "model.add(Conv1D(128, 4, padding='same',))\n",
    "model.add(Activation('relu'))\n",
    "model.add(Dropout(0.4))\n",
    "model.add(MaxPooling1D(pool_size=(2)))\n",
    "model.add(Dropout(0.4))\n",
    "model.add(Flatten())\n",
    "model.add(Dense(len_of_labels))\n",
    "model.add(Activation('softmax'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv1d (Conv1D)              (None, 526, 128)          6784      \n",
      "_________________________________________________________________\n",
      "activation (Activation)      (None, 526, 128)          0         \n",
      "_________________________________________________________________\n",
      "max_pooling1d (MaxPooling1D) (None, 263, 128)          0         \n",
      "_________________________________________________________________\n",
      "conv1d_1 (Conv1D)            (None, 263, 128)          65664     \n",
      "_________________________________________________________________\n",
      "activation_1 (Activation)    (None, 263, 128)          0         \n",
      "_________________________________________________________________\n",
      "dropout (Dropout)            (None, 263, 128)          0         \n",
      "_________________________________________________________________\n",
      "max_pooling1d_1 (MaxPooling1 (None, 131, 128)          0         \n",
      "_________________________________________________________________\n",
      "conv1d_2 (Conv1D)            (None, 131, 128)          65664     \n",
      "_________________________________________________________________\n",
      "activation_2 (Activation)    (None, 131, 128)          0         \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 131, 128)          0         \n",
      "_________________________________________________________________\n",
      "max_pooling1d_2 (MaxPooling1 (None, 65, 128)           0         \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 65, 128)           0         \n",
      "_________________________________________________________________\n",
      "flatten (Flatten)            (None, 8320)              0         \n",
      "_________________________________________________________________\n",
      "dense (Dense)                (None, 8)                 66568     \n",
      "_________________________________________________________________\n",
      "activation_3 (Activation)    (None, 8)                 0         \n",
      "=================================================================\n",
      "Total params: 204,680\n",
      "Trainable params: 204,680\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/40\n",
      "17/17 [==============================] - 2s 120ms/step - loss: 2.1661 - accuracy: 0.2214 - val_loss: 1.7890 - val_accuracy: 0.3774\n",
      "Epoch 2/40\n",
      "17/17 [==============================] - 2s 120ms/step - loss: 1.6110 - accuracy: 0.3814 - val_loss: 1.5725 - val_accuracy: 0.4599\n",
      "Epoch 3/40\n",
      "17/17 [==============================] - 2s 134ms/step - loss: 1.3630 - accuracy: 0.5005 - val_loss: 1.4761 - val_accuracy: 0.4447\n",
      "Epoch 4/40\n",
      "17/17 [==============================] - 3s 149ms/step - loss: 1.2094 - accuracy: 0.5656 - val_loss: 1.4584 - val_accuracy: 0.4729\n",
      "Epoch 5/40\n",
      "17/17 [==============================] - 3s 157ms/step - loss: 1.0609 - accuracy: 0.6130 - val_loss: 1.2971 - val_accuracy: 0.5510\n",
      "Epoch 6/40\n",
      "17/17 [==============================] - 3s 157ms/step - loss: 0.9007 - accuracy: 0.6884 - val_loss: 1.2360 - val_accuracy: 0.5813\n",
      "Epoch 7/40\n",
      "17/17 [==============================] - 2s 120ms/step - loss: 0.8127 - accuracy: 0.7079 - val_loss: 1.2506 - val_accuracy: 0.5748\n",
      "Epoch 8/40\n",
      "17/17 [==============================] - 2s 113ms/step - loss: 0.6349 - accuracy: 0.7851 - val_loss: 1.2724 - val_accuracy: 0.5466\n",
      "Epoch 9/40\n",
      "17/17 [==============================] - 2s 113ms/step - loss: 0.5862 - accuracy: 0.7907 - val_loss: 1.1455 - val_accuracy: 0.5944\n",
      "Epoch 10/40\n",
      "17/17 [==============================] - 2s 127ms/step - loss: 0.4758 - accuracy: 0.8381 - val_loss: 1.2647 - val_accuracy: 0.5618\n",
      "Epoch 11/40\n",
      "17/17 [==============================] - 2s 132ms/step - loss: 0.4333 - accuracy: 0.8512 - val_loss: 1.0637 - val_accuracy: 0.6334\n",
      "Epoch 12/40\n",
      "17/17 [==============================] - 2s 123ms/step - loss: 0.3214 - accuracy: 0.8837 - val_loss: 1.1004 - val_accuracy: 0.6095\n",
      "Epoch 13/40\n",
      "17/17 [==============================] - 2s 120ms/step - loss: 0.3156 - accuracy: 0.8902 - val_loss: 1.1033 - val_accuracy: 0.5965\n",
      "Epoch 14/40\n",
      "17/17 [==============================] - 2s 120ms/step - loss: 0.2357 - accuracy: 0.9237 - val_loss: 1.2402 - val_accuracy: 0.6074\n",
      "Epoch 15/40\n",
      "17/17 [==============================] - 2s 111ms/step - loss: 0.3169 - accuracy: 0.8921 - val_loss: 0.9768 - val_accuracy: 0.6659\n",
      "Epoch 16/40\n",
      "17/17 [==============================] - 2s 114ms/step - loss: 0.1892 - accuracy: 0.9367 - val_loss: 1.0091 - val_accuracy: 0.6291\n",
      "Epoch 17/40\n",
      "17/17 [==============================] - 2s 121ms/step - loss: 0.2320 - accuracy: 0.9135 - val_loss: 1.0554 - val_accuracy: 0.6486\n",
      "Epoch 18/40\n",
      "17/17 [==============================] - 2s 111ms/step - loss: 0.1690 - accuracy: 0.9460 - val_loss: 1.0947 - val_accuracy: 0.6464\n",
      "Epoch 19/40\n",
      "17/17 [==============================] - 2s 113ms/step - loss: 0.1356 - accuracy: 0.9572 - val_loss: 1.0489 - val_accuracy: 0.6421\n",
      "Epoch 20/40\n",
      "17/17 [==============================] - 2s 114ms/step - loss: 0.1445 - accuracy: 0.9526 - val_loss: 1.0328 - val_accuracy: 0.6681\n",
      "Epoch 21/40\n",
      "17/17 [==============================] - 2s 115ms/step - loss: 0.1279 - accuracy: 0.9544 - val_loss: 1.1323 - val_accuracy: 0.6377\n",
      "Epoch 22/40\n",
      "17/17 [==============================] - 2s 115ms/step - loss: 0.1193 - accuracy: 0.9581 - val_loss: 1.1021 - val_accuracy: 0.6334\n",
      "Epoch 23/40\n",
      "17/17 [==============================] - 2s 117ms/step - loss: 0.1210 - accuracy: 0.9572 - val_loss: 1.0572 - val_accuracy: 0.6594\n",
      "Epoch 24/40\n",
      "17/17 [==============================] - 2s 122ms/step - loss: 0.0904 - accuracy: 0.9693 - val_loss: 1.1207 - val_accuracy: 0.6638\n",
      "Epoch 25/40\n",
      "17/17 [==============================] - 2s 116ms/step - loss: 0.1085 - accuracy: 0.9628 - val_loss: 1.1198 - val_accuracy: 0.6551\n",
      "Epoch 26/40\n",
      "17/17 [==============================] - 2s 128ms/step - loss: 0.0600 - accuracy: 0.9795 - val_loss: 1.2624 - val_accuracy: 0.6464\n",
      "Epoch 27/40\n",
      "17/17 [==============================] - 2s 122ms/step - loss: 0.1278 - accuracy: 0.9544 - val_loss: 1.1580 - val_accuracy: 0.6573\n",
      "Epoch 28/40\n",
      "17/17 [==============================] - 2s 126ms/step - loss: 0.0646 - accuracy: 0.9777 - val_loss: 1.1207 - val_accuracy: 0.6703\n",
      "Epoch 29/40\n",
      "17/17 [==============================] - 2s 125ms/step - loss: 0.1013 - accuracy: 0.9702 - val_loss: 1.3129 - val_accuracy: 0.6247\n",
      "Epoch 30/40\n",
      "17/17 [==============================] - 2s 118ms/step - loss: 0.0913 - accuracy: 0.9647 - val_loss: 1.2069 - val_accuracy: 0.6573\n",
      "Epoch 31/40\n",
      "17/17 [==============================] - 2s 116ms/step - loss: 0.0795 - accuracy: 0.9777 - val_loss: 1.2107 - val_accuracy: 0.6659\n",
      "Epoch 32/40\n",
      "17/17 [==============================] - 2s 116ms/step - loss: 0.0978 - accuracy: 0.9693 - val_loss: 1.1536 - val_accuracy: 0.6616\n",
      "Epoch 33/40\n",
      "17/17 [==============================] - 2s 117ms/step - loss: 0.1072 - accuracy: 0.9712 - val_loss: 1.1548 - val_accuracy: 0.6768\n",
      "Epoch 34/40\n",
      "17/17 [==============================] - 2s 128ms/step - loss: 0.0465 - accuracy: 0.9898 - val_loss: 1.2449 - val_accuracy: 0.6573\n",
      "Epoch 35/40\n",
      "17/17 [==============================] - 2s 123ms/step - loss: 0.0550 - accuracy: 0.9898 - val_loss: 1.2430 - val_accuracy: 0.6594\n",
      "Epoch 36/40\n",
      "17/17 [==============================] - 2s 129ms/step - loss: 0.0697 - accuracy: 0.9758 - val_loss: 1.2103 - val_accuracy: 0.6594\n",
      "Epoch 37/40\n",
      "17/17 [==============================] - 2s 123ms/step - loss: 0.0531 - accuracy: 0.9814 - val_loss: 1.2562 - val_accuracy: 0.6876\n",
      "Epoch 38/40\n",
      "17/17 [==============================] - 2s 116ms/step - loss: 0.0754 - accuracy: 0.9702 - val_loss: 1.2241 - val_accuracy: 0.6638\n",
      "Epoch 39/40\n",
      "17/17 [==============================] - 2s 117ms/step - loss: 0.0788 - accuracy: 0.9702 - val_loss: 1.2100 - val_accuracy: 0.6725\n",
      "Epoch 40/40\n",
      "17/17 [==============================] - 2s 118ms/step - loss: 0.0436 - accuracy: 0.9823 - val_loss: 1.3298 - val_accuracy: 0.6508\n"
     ]
    }
   ],
   "source": [
    "from keras.optimizers import RMSprop\n",
    "\n",
    "model.compile(loss= \"categorical_crossentropy\",\n",
    "              optimizer= RMSprop(lr=0.001),\n",
    "              metrics=['accuracy'])\n",
    "\n",
    "cnnhistory= model.fit(X_train, y_train_onehot_encoded,batch_size= 64, epochs= 40,validation_data=(X_test, y_test_onehot_encoded))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "15/15 [==============================] - 0s 11ms/step - loss: 1.3298 - accuracy: 0.6508\n",
      "validation accuracy: 0.650759220123291\n"
     ]
    }
   ],
   "source": [
    "result = model.evaluate(X_test, y_test_onehot_encoded)\n",
    "print(\"validation accuracy:\", result[1])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# plot loss and accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(cnnhistory.history['loss'])\n",
    "plt.plot(cnnhistory.history['val_loss'])\n",
    "plt.title('model loss')\n",
    "plt.ylabel('loss')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'test'], loc='upper left')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(cnnhistory.history['accuracy'])\n",
    "plt.plot(cnnhistory.history['val_accuracy'])\n",
    "plt.title('model accuracy')\n",
    "plt.ylabel('acc')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'test'], loc='upper left')\n",
    "plt.show()"
   ]
  },
  {
   "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": "3.6.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}