exercise.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div dir=\"rtl\">\r\n",
    "\r\n",
    "# <center> به نام خدا </center>\r\n",
    "\r\n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<center>\r\n",
    "\r\n",
    "# تمرین درس پردازش گفتار\r\n",
    "\r\n",
    "## **استاد:** _دکتر بختیاری_\r\n",
    "\r\n",
    "## **دانشجو:** _محمدعلی زرین زاده_\r\n",
    "\r\n",
    "## **تابستان 1400**\r\n",
    "\r\n",
    "</center>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div dir=\"rtl\">\r\n",
    "\r\n",
    "این تمرین با استفاده از `jupyter notebook` پایتون پیاده سازی شده است. هر دو حالت `DTW` و `HMM` انجام شده است.\r\n",
    "\r\n",
    "تمامی داده های مورد استفاده، صورت تمرین و کدها  را می توانید در  [اینجا](https://github.com/MrGolden1/Speech-Processing-Exersice) بیابید.\r\n",
    "\r\n",
    "ابتدا کتابخانه ها و ماژول های لازم را فراخوانی میکنیم.\r\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 130,
   "metadata": {},
   "outputs": [],
   "source": [
    "import librosa\r\n",
    "import librosa.display\r\n",
    "import soundfile\r\n",
    "import matplotlib.pyplot as plt\r\n",
    "import matplotlib\r\n",
    "import numpy as np\r\n",
    "from prettytable import PrettyTable\r\n",
    "from scipy.signal import get_window\r\n",
    "from hmmlearn import hmm\r\n",
    "import glob\r\n",
    "from audiomentations import Compose, AddGaussianNoise, TimeStretch,ClippingDistortion ,AddBackgroundNoise,Gain,PolarityInversion\r\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div dir=\"rtl\">\r\n",
    "\r\n",
    "## داده های ما از 4 بخش زیر تشکیل شده است: \r\n",
    "\r\n",
    "1. ده فایل صوتی نام خانوادگی برای آموزش \r\n",
    "\r\n",
    "2. سه فایل نام خانوادگی برای تست\r\n",
    "\r\n",
    "3. سه فایل نام خانوادگی با صدای فردی دیگر\r\n",
    "\r\n",
    "4. سه فایل نام \r\n",
    " \r\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 143,
   "metadata": {},
   "outputs": [],
   "source": [
    "lastname_train_path = 'lastname_train/'\r\n",
    "lastname_test_path = 'lastname_test/'\r\n",
    "lastname_another_path = 'lastname_another_person/'\r\n",
    "first_name_path = 'firstname/'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div dir=\"rtl\">\r\n",
    "\r\n",
    "ابتدا هر بخش از داده ها را بارگیری کرده و درون لیست های جدا قرار میدهیم. تنها پیش پردازشی که همزمان با بارگیری صورت میگیرد حذف سکوت با استفاده از تابع `librosa.effects.trim` می باشد.\r\n",
    "\r\n",
    "\r\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "lastname_train_files = glob.glob(lastname_train_path + '*.wav')\r\n",
    "lastname_trains = []\r\n",
    "\r\n",
    "for path in lastname_train_files:\r\n",
    "    y, sr = librosa.load(path, sr=11025,mono=True)\r\n",
    "    y, index = librosa.effects.trim(y,top_db=10)\r\n",
    "    librosa.display.waveshow(y)\r\n",
    "    lastname_trains.append([y, sr])\r\n",
    "    \r\n",
    "train_size = len(lastname_trains)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "lastname_test_files = glob.glob(lastname_test_path + '*.wav')\r\n",
    "lastname_tests = []\r\n",
    "\r\n",
    "for path in lastname_test_files:\r\n",
    "    y, sr = librosa.load(path, sr=11025,mono=True)\r\n",
    "    y, index = librosa.effects.trim(y,top_db=10)\r\n",
    "    librosa.display.waveshow(y)\r\n",
    "    lastname_tests.append([y, sr])\r\n",
    "    \r\n",
    "test_size = len(lastname_tests)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "firstname_files = glob.glob(first_name_path + '*.wav')\r\n",
    "firstnames = []\r\n",
    "\r\n",
    "for path in firstname_files:\r\n",
    "    y, sr = librosa.load(path, sr=11025,mono=True)\r\n",
    "    y, index = librosa.effects.trim(y,top_db=10)\r\n",
    "    librosa.display.waveshow(y)\r\n",
    "    firstnames.append([y, sr])\r\n",
    "    \r\n",
    "firstname_size = len(firstnames)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 147,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "lastname_another_files = glob.glob(lastname_another_path + '*.wav')\r\n",
    "lastname_anothers = []\r\n",
    "\r\n",
    "for path in lastname_another_files:\r\n",
    "    y, sr = librosa.load(path, sr=11025,mono=True)\r\n",
    "    y, index = librosa.effects.trim(y,top_db=10)\r\n",
    "    librosa.display.waveshow(y)\r\n",
    "    lastname_anothers.append([y, sr])\r\n",
    "    \r\n",
    "lastname_another_size = len(lastname_anothers)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div dir=\"rtl\">\r\n",
    "\r\n",
    "در اولین مرحله فاصله ی نام های خانوادگی را با یکدیگر در حالت خام می سنجیم. برای این کار از  تابع `librosa.sequence.dtw` استفاده میکنیم. خروجی این تابع شامل دو آرایه (یکی ماتریس هزینه و دیگری ماتریس مسیر بهینه) می باشد. کمترین فاصله در آخرین سطر و ستون ماتریس هزینه قرار دارد.\r\n",
    "\r\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [],
   "source": [
    "\r\n",
    "distances_lastname_lastname = np.zeros((train_size, train_size))\r\n",
    "for i in range(train_size):\r\n",
    "    for j in range(i+1, train_size):\r\n",
    "        x_1, fs = lastname_trains[i]\r\n",
    "        x_2, fs = lastname_trains[j]\r\n",
    "        D, wp = librosa.sequence.dtw(x_1, x_2)\r\n",
    "        best_cost = D[-1, -1]\r\n",
    "        distances_lastname_lastname[i][j] = round(best_cost, 3)\r\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div dir=\"rtl\">\r\n",
    "\r\n",
    "نتیجه را در جدول زیر مشاهده می نمایید. \r\n",
    "\r\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+---+-----+---------+---------+---------+---------+---------+---------+---------+---------+---------+\n",
      "|   |  0  |    1    |    2    |    3    |    4    |    5    |    6    |    7    |    8    |    9    |\n",
      "+---+-----+---------+---------+---------+---------+---------+---------+---------+---------+---------+\n",
      "| 0 | 0.0 | 362.079 |  285.01 | 346.856 | 338.453 | 418.572 | 479.444 | 328.333 | 339.433 | 306.194 |\n",
      "| 1 | 0.0 |   0.0   | 292.001 | 334.531 | 280.369 | 343.209 | 365.764 | 317.607 | 366.611 |  274.61 |\n",
      "| 2 | 0.0 |   0.0   |   0.0   | 330.453 |  305.4  | 360.345 | 433.734 | 302.903 | 276.518 | 256.093 |\n",
      "| 3 | 0.0 |   0.0   |   0.0   |   0.0   | 309.842 | 322.802 | 348.941 | 311.924 | 385.655 | 326.122 |\n",
      "| 4 | 0.0 |   0.0   |   0.0   |   0.0   |   0.0   | 333.417 | 332.407 | 267.814 | 381.062 | 286.046 |\n",
      "| 5 | 0.0 |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   | 394.436 | 387.106 | 366.614 | 346.852 |\n",
      "| 6 | 0.0 |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   | 316.297 | 505.957 | 395.449 |\n",
      "| 7 | 0.0 |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   | 402.075 | 307.212 |\n",
      "| 8 | 0.0 |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   | 310.624 |\n",
      "| 9 | 0.0 |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   |\n",
      "+---+-----+---------+---------+---------+---------+---------+---------+---------+---------+---------+\n"
     ]
    }
   ],
   "source": [
    "labels = [str(i) for i in range(train_size)]\r\n",
    "pretty_table = PrettyTable([''] + labels)\r\n",
    "for i in range(train_size):\r\n",
    "    pretty_table.add_row([i]+ distances_lastname_lastname[i,:].tolist())\r\n",
    "print(pretty_table.get_string())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div dir=\"rtl\">\r\n",
    "\r\n",
    "حال در ادامه فاصله نام خانوداگی و نام را در حالت داده خام با همان الگوریتم بررسی میکنیم.\r\n",
    "\r\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 163,
   "metadata": {},
   "outputs": [],
   "source": [
    "distances_firstname_lastname = np.zeros((train_size, firstname_size))\r\n",
    "for i in range(train_size):\r\n",
    "    for j in range(firstname_size):\r\n",
    "        x_1, fs = lastname_trains[i]\r\n",
    "        x_2, fs = firstnames[j]\r\n",
    "        D, wp = librosa.sequence.dtw(x_1, x_2)\r\n",
    "        best_cost = D[-1, -1]\r\n",
    "        distances_firstname_lastname[i][j] = round(best_cost, 3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div dir=\"rtl\">\r\n",
    "\r\n",
    "همانطور که مشاهده می شود به طور میانگین اختلاف نام با نام خانوادگی بیشتر از دو نمونه از نام خانوادگی می باشد. اما آنچنان تمایز قابل توجهی ایجاد نمی کند.\r\n",
    "</div>\r\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 164,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+---+---------+---------+---------+\n",
      "|   |    0    |    1    |    2    |\n",
      "+---+---------+---------+---------+\n",
      "| 0 | 362.137 | 400.997 | 406.274 |\n",
      "| 1 | 460.296 | 475.236 | 467.854 |\n",
      "| 2 | 405.668 | 431.357 | 435.792 |\n",
      "| 3 | 495.076 | 506.675 | 490.857 |\n",
      "| 4 |  450.59 | 463.808 | 479.196 |\n",
      "| 5 |  597.7  | 603.347 | 562.326 |\n",
      "| 6 | 551.504 | 551.316 | 553.147 |\n",
      "| 7 | 418.793 |  434.91 | 451.157 |\n",
      "| 8 | 470.139 | 487.566 | 467.337 |\n",
      "| 9 | 423.306 | 448.694 | 429.574 |\n",
      "+---+---------+---------+---------+\n"
     ]
    }
   ],
   "source": [
    "labels = [str(i) for i in range(firstname_size)]\r\n",
    "pretty_table = PrettyTable([''] + labels)\r\n",
    "for i in range(train_size):\r\n",
    "    pretty_table.add_row([i]+ distances_firstname_lastname[i,:].tolist())\r\n",
    "print(pretty_table.get_string())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div dir=\"rtl\">\r\n",
    "\r\n",
    "در آخرین پردازش با داده های خام، نام خانوادگی های ضبط شده توسط دو گوینده متفاوت را با یکدیگر مقایسه می کنیم.\r\n",
    "\r\n",
    "</div>\r\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 155,
   "metadata": {},
   "outputs": [],
   "source": [
    "distances_firstname_lastname_lastname_another = np.zeros((train_size, lastname_another_size))\r\n",
    "for i in range(train_size):\r\n",
    "    for j in range(lastname_another_size):\r\n",
    "        x_1, fs = lastname_trains[i]\r\n",
    "        x_2, fs = lastname_anothers[j]\r\n",
    "        D, wp = librosa.sequence.dtw(x_1, x_2)\r\n",
    "        best_cost = D[-1, -1]\r\n",
    "        distances_firstname_lastname_lastname_another[i][j] = round(best_cost, 3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div dir=\"rtl\">\r\n",
    "\r\n",
    "نتیجه نشان از بیشتر بودن فاصله نمونه ها نسبت به حالتی که فقط یک گوینده داشتیم، دارد.\r\n",
    "\r\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 156,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+---+---------+---------+---------+\n",
      "|   |    0    |    1    |    2    |\n",
      "+---+---------+---------+---------+\n",
      "| 0 | 336.635 |  400.75 |  370.68 |\n",
      "| 1 | 481.834 | 551.255 | 448.792 |\n",
      "| 2 | 416.754 | 474.771 | 377.543 |\n",
      "| 3 |  543.15 | 634.567 | 446.498 |\n",
      "| 4 |  465.13 | 531.959 | 459.695 |\n",
      "| 5 | 645.091 | 728.822 |  496.89 |\n",
      "| 6 | 606.269 | 701.109 | 528.129 |\n",
      "| 7 | 410.444 | 487.915 | 436.848 |\n",
      "| 8 |  482.91 |  550.82 | 391.545 |\n",
      "| 9 | 438.855 | 513.495 |  401.01 |\n",
      "+---+---------+---------+---------+\n"
     ]
    }
   ],
   "source": [
    "labels = [str(i) for i in range(lastname_another_size)]\r\n",
    "pretty_table = PrettyTable([''] + labels)\r\n",
    "for i in range(train_size):\r\n",
    "    pretty_table.add_row([i]+ distances_firstname_lastname_lastname_another[i,:].tolist())\r\n",
    "print(pretty_table.get_string())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div dir=\"rtl\">\r\n",
    "\r\n",
    "بدیهی است که در حوزه پردازش گفتار، از داده های خام هیچ الگو و قاعده ای را نمی توان شناسایی نمود. استخراج ویژگی امری اجتناب ناپذیر است. در ادامه دو پیش پردازش انجام می شود. یکی تبدیل هر نمونه به فریم هایی با طول برابر و دارای همپوشانی و دیگری استخراج ویژگی از هر یک از آن فریم ها می باشد.\r\n",
    "\r\n",
    "برای استخراج ویژگی موارد زیر در نظر گرفته شده است:\r\n",
    "\r\n",
    "* 13 ضریب اول MFCC\r\n",
    "\r\n",
    "* مشتق ویژگی اول (شامل 13 عدد)\r\n",
    "\r\n",
    "* مشتق مرتبه دوم ویژگی اول (شامل 13 عدد)\r\n",
    "\r\n",
    "* نرخ عبور از صفر\r\n",
    "\r\n",
    "* انرژی سیگنال\r\n",
    "\r\n",
    "در مجموع هر نمونه ما از تعداد متفاوتی فریم و هر فریم دقیقا از 41 ویژگی تشکیل میشود. برای محاسبه وِیژگی های ذکر شده از توابع `librosa.feature.mfcc`، `librosa.feature.zero_crossing_rate` و ` librosa.feature.rms` استفاده شده است. سایز هر پنجره 50 میلی ثانیه و همپوشانی 10 میلی ثانیه در نظر گرفته شده است. در ادامه همه داده ها و نمونه های خام را به وِیژگی های فوق تبدیل می کنیم:\r\n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## hop: _number of bits between two succesive window_\r\n",
    "\r\n",
    "## n_ftt: _number of bits in one frame_\r\n",
    "\r\n",
    "## overlap: _nftt - hop_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "frame_length = int(0.05*sr)\r\n",
    "hop_length = int(0.04*sr)\r\n",
    "lastname_trains_features = []\r\n",
    "for y, fs in lastname_trains:\r\n",
    "    mfcc = librosa.feature.mfcc(y, fs, n_mfcc=13, hop_length=hop_length, n_fft=frame_length, window=get_window('hamming',frame_length))\r\n",
    "    mfcc_delta = librosa.feature.delta(mfcc)\r\n",
    "    mfcc_delta2 = librosa.feature.delta(mfcc,order=2)\r\n",
    "    zcr = librosa.feature.zero_crossing_rate(y,frame_length=frame_length,hop_length=hop_length)\r\n",
    "    rms = librosa.feature.rms(y,frame_length=frame_length,hop_length=hop_length)\r\n",
    "    features = np.concatenate([mfcc,mfcc_delta,mfcc_delta2,zcr,rms])\r\n",
    "    lastname_trains_features.append(features)\r\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div dir=\"rtl\">\r\n",
    "\r\n",
    "# نرمال سازی\r\n",
    "\r\n",
    "نکته مهم بعد از استخراج وِیژگی در تمامی سیستم های یادگیری ماشین و شناسایی الگو، هم مقیاس کردن تمامی ویژگی ها می باشد. برای این مورد ما از نورمال سازی رایج میانگین-واریانس استفاده کرده ایم. \r\n",
    "\r\n",
    "</div>\r\n",
    "\r\n",
    "$$ x^{norm} = \\frac{x - mean}{stdev} $$ "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "def cmvn(vecs, variance_normalization=False):\r\n",
    "    \"\"\" This function is aimed to perform global cepstral mean and\r\n",
    "        variance normalization (CMVN) on input feature vector \"vec\".\r\n",
    "        The code assumes that there is one observation per row.\r\n",
    "    Args:\r\n",
    "        vecs  (list[array]): list of input feature matrises\r\n",
    "            (array size:(num_features,num_observation))\r\n",
    "        variance_normalization (bool): If the variance\r\n",
    "            normilization should be performed or not.\r\n",
    "    Return:\r\n",
    "          array: The mean(or mean+variance) normalized feature vector.\r\n",
    "    \"\"\"\r\n",
    "\r\n",
    "    # concarinate all matrises to one\r\n",
    "    temp_vec = np.concatenate(vecs, axis=1)\r\n",
    "\r\n",
    "    eps = 2**-30\r\n",
    "    rows, cols = temp_vec.shape\r\n",
    "\r\n",
    "    # Mean calculation\r\n",
    "    mean = np.mean(temp_vec, axis=1, keepdims=True)\r\n",
    "    mean_vec = np.tile(mean, (1, cols))\r\n",
    "\r\n",
    "    norm_vecs = []\r\n",
    "    # Variance normalization\r\n",
    "    if variance_normalization:\r\n",
    "        # Mean subtraction\r\n",
    "        mean_subtracted = temp_vec - mean_vec\r\n",
    "        stdev = np.std(mean_subtracted, axis=1, keepdims=True)\r\n",
    "\r\n",
    "        for item in vecs:\r\n",
    "            norm_vecs.append(\r\n",
    "                (item - np.tile(mean, (1, item.shape[1]))) / (np.tile(stdev, (1, item.shape[1])) + eps))\r\n",
    "\r\n",
    "    else:\r\n",
    "        for item in vecs:\r\n",
    "            norm_vecs.append(\r\n",
    "                (item - np.tile(mean, (1, item.shape[1]))))\r\n",
    "\r\n",
    "    return norm_vecs\r\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "normalized_lastname_trains_features = cmvn(lastname_trains_features, True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div dir=\"rtl\">\r\n",
    "\r\n",
    "طیف ضرایب MFCC و مشتق مرتبه اول و دوم آن را برای یکی از نمونه ها پس از نورمال سازی مشاهده میکنید.\r\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.colorbar.Colorbar at 0x2901f9a5808>"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 6 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots(nrows=3, sharex=True, sharey=True)\r\n",
    "img1 = librosa.display.specshow(mfcc, ax=ax[0], x_axis='time')\r\n",
    "ax[0].set(title='MFCC')\r\n",
    "ax[0].label_outer()\r\n",
    "img2 = librosa.display.specshow(mfcc_delta, ax=ax[1], x_axis='time')\r\n",
    "ax[1].set(title=r'MFCC-$\\Delta$')\r\n",
    "ax[1].label_outer()\r\n",
    "img3 = librosa.display.specshow(mfcc_delta2, ax=ax[2], x_axis='time')\r\n",
    "ax[2].set(title=r'MFCC-$\\Delta^2$')\r\n",
    "fig.colorbar(img1, ax=[ax[0]])\r\n",
    "fig.colorbar(img2, ax=[ax[1]])\r\n",
    "fig.colorbar(img3, ax=[ax[2]])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div dir=\"rtl\">\r\n",
    "\r\n",
    "در ادامه ویژگی های دیگر داده هایمان را  استخراج شده و آن ها را نورمال میکنیم.\r\n",
    "\r\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [],
   "source": [
    "frame_length = int(0.05*sr)\r\n",
    "hop_length = int(0.04*sr)\r\n",
    "lastname_tests_features = []\r\n",
    "for y, fs in lastname_tests:\r\n",
    "    mfcc = librosa.feature.mfcc(y, fs, n_mfcc=13, hop_length=hop_length, n_fft=frame_length, window=get_window('hamming',frame_length))\r\n",
    "    mfcc_delta = librosa.feature.delta(mfcc)\r\n",
    "    mfcc_delta2 = librosa.feature.delta(mfcc,order=2)\r\n",
    "    zcr = librosa.feature.zero_crossing_rate(y,frame_length=frame_length,hop_length=hop_length)\r\n",
    "    rms = librosa.feature.rms(y,frame_length=frame_length,hop_length=hop_length)\r\n",
    "    features = np.concatenate([mfcc,mfcc_delta,mfcc_delta2,zcr,rms])\r\n",
    "    lastname_tests_features.append(features)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [],
   "source": [
    "frame_length = int(0.05*sr)\r\n",
    "hop_length = int(0.04*sr)\r\n",
    "firstname_features = []\r\n",
    "for y, fs in firstnames:\r\n",
    "    mfcc = librosa.feature.mfcc(y, fs, n_mfcc=13, hop_length=hop_length, n_fft=frame_length, window=get_window('hamming',frame_length))\r\n",
    "    mfcc_delta = librosa.feature.delta(mfcc)\r\n",
    "    mfcc_delta2 = librosa.feature.delta(mfcc,order=2)\r\n",
    "    zcr = librosa.feature.zero_crossing_rate(y,frame_length=frame_length,hop_length=hop_length)\r\n",
    "    rms = librosa.feature.rms(y,frame_length=frame_length,hop_length=hop_length)\r\n",
    "    features = np.concatenate([mfcc,mfcc_delta,mfcc_delta2,zcr,rms])\r\n",
    "    firstname_features.append(features)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [],
   "source": [
    "normalized_firstnames_features = cmvn(firstname_features,True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [],
   "source": [
    "\r\n",
    "distances_with_features_lastname = np.zeros((train_size, train_size))\r\n",
    "for i in range(train_size):\r\n",
    "    for j in range(i+1, train_size):\r\n",
    "        x_1 = normalized_lastname_trains_features[i]\r\n",
    "        x_2 = normalized_lastname_trains_features[j]\r\n",
    "        D, wp = librosa.sequence.dtw(x_1, x_2)\r\n",
    "        best_cost = D[-1, -1]\r\n",
    "        distances_with_features_lastname[i][j] = round(best_cost, 3)\r\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 145,
   "metadata": {},
   "outputs": [],
   "source": [
    "frame_length = int(0.05*sr)\r\n",
    "hop_length = int(0.04*sr)\r\n",
    "lastname_another_features = []\r\n",
    "for y, fs in lastname_anothers:\r\n",
    "    mfcc = librosa.feature.mfcc(y, fs, n_mfcc=13, hop_length=hop_length, n_fft=frame_length, window=get_window('hamming',frame_length))\r\n",
    "    mfcc_delta = librosa.feature.delta(mfcc)\r\n",
    "    mfcc_delta2 = librosa.feature.delta(mfcc,order=2)\r\n",
    "    zcr = librosa.feature.zero_crossing_rate(y,frame_length=frame_length,hop_length=hop_length)\r\n",
    "    rms = librosa.feature.rms(y,frame_length=frame_length,hop_length=hop_length)\r\n",
    "    features = np.concatenate([mfcc,mfcc_delta,mfcc_delta2,zcr,rms])\r\n",
    "    lastname_another_features.append(features)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 146,
   "metadata": {},
   "outputs": [],
   "source": [
    "normalized_lastname_another_features = cmvn(lastname_another_features,True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+---+-----+---------+---------+---------+---------+---------+---------+---------+---------+---------+\n",
      "|   |  0  |    1    |    2    |    3    |    4    |    5    |    6    |    7    |    8    |    9    |\n",
      "+---+-----+---------+---------+---------+---------+---------+---------+---------+---------+---------+\n",
      "| 0 | 0.0 | 180.924 | 137.502 | 141.066 | 175.856 |  148.02 |  186.84 | 169.114 | 172.871 | 162.052 |\n",
      "| 1 | 0.0 |   0.0   | 142.995 | 172.745 | 152.051 | 143.468 | 164.853 | 156.218 |  165.29 | 140.204 |\n",
      "| 2 | 0.0 |   0.0   |   0.0   | 137.175 | 141.992 | 126.719 | 167.706 | 160.458 | 130.575 | 125.739 |\n",
      "| 3 | 0.0 |   0.0   |   0.0   |   0.0   | 155.472 | 144.386 | 152.998 | 146.428 | 171.451 | 157.715 |\n",
      "| 4 | 0.0 |   0.0   |   0.0   |   0.0   |   0.0   |  144.08 | 135.552 | 131.027 | 170.491 |  146.05 |\n",
      "| 5 | 0.0 |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   | 158.272 | 151.321 | 142.445 | 138.324 |\n",
      "| 6 | 0.0 |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   | 125.931 | 205.013 | 170.287 |\n",
      "| 7 | 0.0 |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   | 195.742 | 155.229 |\n",
      "| 8 | 0.0 |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   | 145.777 |\n",
      "| 9 | 0.0 |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   |\n",
      "+---+-----+---------+---------+---------+---------+---------+---------+---------+---------+---------+\n"
     ]
    }
   ],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div dir=\"rtl\">\r\n",
    "\r\n",
    "حال همه ی فاصله هایی که در داده های خام محاسبه کردیم را دوباره برای داده های با ویژگی نورمال شده تکرار میکنیم.\r\n",
    "\r\n",
    "فاصله ده نام خانوادگی آموزشی:\r\n",
    "\r\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+---+-----+---------+---------+---------+---------+---------+---------+---------+---------+---------+\n",
      "|   |  0  |    1    |    2    |    3    |    4    |    5    |    6    |    7    |    8    |    9    |\n",
      "+---+-----+---------+---------+---------+---------+---------+---------+---------+---------+---------+\n",
      "| 0 | 0.0 | 180.924 | 137.502 | 141.066 | 175.856 |  148.02 |  186.84 | 169.114 | 172.871 | 162.052 |\n",
      "| 1 | 0.0 |   0.0   | 142.995 | 172.745 | 152.051 | 143.468 | 164.853 | 156.218 |  165.29 | 140.204 |\n",
      "| 2 | 0.0 |   0.0   |   0.0   | 137.175 | 141.992 | 126.719 | 167.706 | 160.458 | 130.575 | 125.739 |\n",
      "| 3 | 0.0 |   0.0   |   0.0   |   0.0   | 155.472 | 144.386 | 152.998 | 146.428 | 171.451 | 157.715 |\n",
      "| 4 | 0.0 |   0.0   |   0.0   |   0.0   |   0.0   |  144.08 | 135.552 | 131.027 | 170.491 |  146.05 |\n",
      "| 5 | 0.0 |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   | 158.272 | 151.321 | 142.445 | 138.324 |\n",
      "| 6 | 0.0 |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   | 125.931 | 205.013 | 170.287 |\n",
      "| 7 | 0.0 |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   | 195.742 | 155.229 |\n",
      "| 8 | 0.0 |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   | 145.777 |\n",
      "| 9 | 0.0 |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   |   0.0   |\n",
      "+---+-----+---------+---------+---------+---------+---------+---------+---------+---------+---------+\n"
     ]
    }
   ],
   "source": [
    "labels = [str(i) for i in range(train_size)]\r\n",
    "pretty_table = PrettyTable([''] + labels)\r\n",
    "for i in range(train_size):\r\n",
    "    pretty_table.add_row([i]+ distances_with_features_lastname[i,:].tolist())\r\n",
    "print(pretty_table.get_string())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [],
   "source": [
    "labels = [str(i) for i in range(firstname_size)]\r\n",
    "pretty_table = PrettyTable([''] + labels)\r\n",
    "for i in range(train_size):\r\n",
    "    pretty_table.add_row([i]+ distances_with_features_firstname_lastname[i,:].tolist())\r\n",
    "print(pretty_table.get_string())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div dir=\"rtl\">\r\n",
    "\r\n",
    "فاصله ده نام خانوادگی آموزشی با نام ها:\r\n",
    "\r\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+---+---------+---------+---------+\n",
      "|   |    0    |    1    |    2    |\n",
      "+---+---------+---------+---------+\n",
      "| 0 | 247.887 |  242.55 | 246.219 |\n",
      "| 1 | 263.637 | 266.646 | 259.557 |\n",
      "| 2 |  239.03 | 237.599 | 235.467 |\n",
      "| 3 | 254.367 | 250.646 | 246.967 |\n",
      "| 4 |  259.72 | 259.826 | 258.732 |\n",
      "| 5 | 238.676 |  239.32 | 237.603 |\n",
      "| 6 | 276.008 | 279.783 | 277.591 |\n",
      "| 7 | 256.789 | 255.418 | 255.486 |\n",
      "| 8 | 209.354 | 209.033 |  201.57 |\n",
      "| 9 | 229.366 | 230.135 | 223.854 |\n",
      "+---+---------+---------+---------+\n"
     ]
    }
   ],
   "source": [
    "X = np.concatenate([np.transpose(item) for item in normalized_lastname_trains_features])\r\n",
    "n_components = 5\r\n",
    "lengths = np.array([item.shape[1] for item in normalized_lastname_trains_features])\r\n",
    "hmm_model = hmm.GaussianHMM(\r\n",
    "    n_components=n_components, covariance_type=\"full\", n_iter=100)\r\n",
    "hmm_model.fit(X,lengths)\r\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 157,
   "metadata": {},
   "outputs": [],
   "source": [
    "distances_with_features_firstname_lastname_lastname_another = np.zeros((train_size, lastname_another_size))\r\n",
    "for i in range(train_size):\r\n",
    "    for j in range(lastname_another_size):\r\n",
    "        x_1 = normalized_lastname_trains_features[i]\r\n",
    "        x_2 = normalized_lastname_another_features[j]\r\n",
    "        D, wp = librosa.sequence.dtw(x_1, x_2)\r\n",
    "        best_cost = D[-1, -1]\r\n",
    "        distances_with_features_firstname_lastname_lastname_another[i][j] = round(best_cost, 3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div dir=\"rtl\">\r\n",
    "\r\n",
    "فاصله ده نام خانوادگی آموزشی با نام خانوادگی ضبط شده توسط گوینده ای دیگر:\r\n",
    "\r\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 158,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+---+---------+---------+---------+\n",
      "|   |    0    |    1    |    2    |\n",
      "+---+---------+---------+---------+\n",
      "| 0 | 200.608 | 196.946 | 284.247 |\n",
      "| 1 |  189.37 | 179.002 | 286.719 |\n",
      "| 2 | 165.718 | 164.351 | 259.797 |\n",
      "| 3 | 199.396 | 193.922 | 300.305 |\n",
      "| 4 | 198.447 | 182.123 | 290.444 |\n",
      "| 5 | 182.289 | 178.708 |  263.52 |\n",
      "| 6 | 214.277 | 206.636 | 307.177 |\n",
      "| 7 | 198.828 |  181.19 | 290.781 |\n",
      "| 8 | 173.561 | 160.685 | 246.003 |\n",
      "| 9 | 162.802 | 165.214 |  255.95 |\n",
      "+---+---------+---------+---------+\n"
     ]
    }
   ],
   "source": [
    "labels = [str(i) for i in range(lastname_another_size)]\r\n",
    "pretty_table = PrettyTable([''] + labels)\r\n",
    "for i in range(train_size):\r\n",
    "    pretty_table.add_row([i]+ distances_with_features_firstname_lastname_lastname_another[i,:].tolist())\r\n",
    "print(pretty_table.get_string())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div dir=\"rtl\">\r\n",
    "\r\n",
    "## تحلیل سه جدول فوق\r\n",
    "\r\n",
    "اگر میانگین و واریانس سه جدول را مقایسه کنیم در میابیم که فواصل به اینگونه می باشد:\r\n",
    "\r\n",
    "`نام خانوداگی توسط یک گوینده` *<* `نام خانوادگی توسط دو گوینده` *<* `نام و نام خانوداگی`\r\n",
    "\r\n",
    "پیشرفت قابل توجه نسبت به فاصله نمونه ها در حالت خام. همچنین اختلاف اعداد درون هر جدول از میانگین بسیار کم تر شده و اعداد به یکدیگر نزدیک هستند.\r\n",
    "\r\n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div dir=\"rtl\">\r\n",
    "\r\n",
    " # <center> آموزش با مدل مخفی مارکو (HMM)  </center>\r\n",
    "\r\n",
    "\r\n",
    "یکی ا ز مدل های رایج برای بررسی پدیده های گفتاری استفاده از مدل های مخفی مارکو می باشد. این مدل ها از نظریات احتمالات و آمار برای بررسی رفتار و مشاهدات متوالی استفاده میکنند. از آنجایی که گفتار هم یک نوع داده متوالی حاصل از یک سیستم مخفی (مجرای گفتار انسان) است می توان از آن استفاده نمود. در پایتون کتابخانه ی `hmmlearn` را برای این منظور به کار میبریم.\r\n",
    "\r\n",
    "مدل ما از 5 جز (state) تشکیل شده است و از 10 فایل نام خانوادگی برای آموزش آن استفاده میکنیم.\r\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 159,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "GaussianHMM(covariance_type='full', n_components=5, n_iter=100)"
      ]
     },
     "execution_count": 159,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "for item in normalized_lastname_trains_features:\r\n",
    "    print(hmm_model.score(np.transpose(item)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div dir=\"rtl\">\r\n",
    "\r\n",
    "لازم به ذکر است که برای ارزیابی و میزان مطابفت مشاهدات جدید با مدل آموزش دیده از تابع `score` استفاده میکنیم که خروجی آن احتمال لگاریتمی در بازه ی زیر می باشد:\r\n",
    "\r\n",
    "</div>\r\n",
    "\r\n",
    "$$ (-\\infty, 0]  $$\r\n",
    "\r\n",
    "<div dir=\"rtl\">\r\n",
    "\r\n",
    "هر میزان که مقدار آن به صفر نزدیک تر باشد احتمال نیز بیشتر است. [اطلاعات بیشتر](https://en.wikipedia.org/wiki/Log_probability)\r\n",
    "\r\n",
    "ابتدا میزان تطابق را برای همان داده هایی که مدل با آن آموزش دیده است محاسبه میکنیم. \r\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [],
   "source": [
    "normalized_lastname_tests_features = cmvn(lastname_tests_features,True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div dir=\"rtl\">\r\n",
    "\r\n",
    "برای نام خانوادگی تست (گوینده یکسان با داده های آموزشی)\r\n",
    "\r\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-5053.7475584087015\n",
      "-7255.334544822906\n",
      "-4537.286386583172\n"
     ]
    }
   ],
   "source": [
    "for item in normalized_lastname_tests_features:\r\n",
    "    print(hmm_model.score(np.transpose(item)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div dir=\"rtl\">\r\n",
    "\r\n",
    "برای نام ها\r\n",
    "\r\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-6512.018373262616\n",
      "-7061.515468930111\n",
      "-6273.59090235505\n"
     ]
    }
   ],
   "source": [
    "for item in normalized_firstnames_features:\r\n",
    "    print(hmm_model.score(np.transpose(item)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div dir=\"rtl\">\r\n",
    "\r\n",
    "برای نام خانوادگی توسط گوینده ای دیگر\r\n",
    "\r\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 160,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-6147.747596445928\n",
      "-5765.352023834199\n",
      "-10327.419936321585\n"
     ]
    }
   ],
   "source": [
    "for item in normalized_lastname_another_features:\r\n",
    "    print(hmm_model.score(np.transpose(item)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div dir=\"rtl\">\r\n",
    "\r\n",
    "در تحلیل نتایج فوق می توان گفت که که مدل ما از مشکل `underfitting` رنج می برد که برای آموزش چنین مدل پیچیده ای صرفا با ده داده طبیعی به نظر میرسد. \r\n",
    "میزان احتمال برای نام خانوادگی تست و نام تمایزی را ایجاد نمیکند. برای حل مشکل فوق یک راه حل جمع آوری و افزودن داده بیشتر در مرحله آموزش است. تنوع در مجموعه داده نیز بسیار مهم است. وقتی کارایی سیستم افزایش پیدا میکند که گفتار های ما در محیط های مختلف، با انواع میکروفن، با گوینده های مختلف و .... جمع آوری شده باشد. این کار مفید اما زمان بر ، وقت گیر و پر هزینه است. به جای آن ما در اینجا از تکینیک دیگر به نام `data augmentation` استفاده نموده ایم. با استفاده از این تکنیک سعی میکنیم با همان مجموعه داده ی محدودی که در اختیار داریم، تعداد و تنوع مورد نظرمان را ایجاد نماییم. \r\n",
    "\r\n",
    "افزونگی هایی که در اینجا انتخاب شده به شرح زیر است:\r\n",
    "\r\n",
    "- **نویز گوسی:** به صورت تصادفی و بر اساس توزیع نورمال مقادیر داده ها را تغییر میدهد. این افزونگی فارق از ماهیت داده عمل میکند.\r\n",
    "- **کشش زمانی:** داده را به صورت زمانی کشیده یا فشرده میکنیم. نکته مهم در این افزونگی آن است که میزان فرکانس نمونه برداری را تغییر نمیدهد.\r\n",
    "- **ClippingDistortion:** بعضی نقاط کمتر یا بیشتر از یک حد آستانه تصادفی بر اساس توزیع یکنواخت را حذف می نماید.\r\n",
    "- **افزودن پس زمینه:** افزودن صداهای محیط های مختلف مانند یک مهمانی ، خیابان ، باران و ... به پس زمینه داده های آموزشی .\r\n",
    "- **Gain:** تغییر بلندی و حجم صدا به صورت تصادفی.\r\n",
    "- **PolarityInversion:** داده های ما نسبت به محور افقی برعکس میشوند. مانند اینکه در 1- ضرب شوند. این افزونگی در صدایی که ما می شنویم تغییری ایجاد نمیکند.\r\n",
    "\r\n",
    "برای اعمال این افزونگی ها از کتابخانه ی `audiomentations` استفاده میکنیم. هر یک از این موارد با احتمال 50 درصد ممکن است بر روی نمونه ها انجام شود یا نشود. \r\n",
    "با این کار 10 داده ی آموزشی خود را به 100 نمونه تبدیل میکنیم. \r\n",
    "برای صدا های پس زمینه از مجموعه داده [freesound](https://annotator.freesound.org/fsd/explore/%252Fm%252F093_4n/) بهره برده ایم.\r\n",
    "\r\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "metadata": {},
   "outputs": [],
   "source": [
    "augment = Compose([\r\n",
    "    AddGaussianNoise(min_amplitude=0.001, max_amplitude=0.015, p=0.5),\r\n",
    "    TimeStretch(min_rate=0.8, max_rate=1.25, p=0.5),\r\n",
    "    ClippingDistortion(p=0.5),\r\n",
    "    AddBackgroundNoise(sounds_path=\"background_noises/\",p=0.5),\r\n",
    "    Gain(p=0.5),\r\n",
    "    PolarityInversion(p=0.5),\r\n",
    "])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "aug_samples = []\r\n",
    "aut_path = \"aug_trains/\"\r\n",
    "for i in range(10):\r\n",
    "    j = 0\r\n",
    "    for y, sr in lastname_trains:\r\n",
    "        aug_sample = augment(\r\n",
    "            samples=y, sample_rate=sr)\r\n",
    "        aug_samples.append([aug_sample,sr])\r\n",
    "        soundfile.write(aut_path + \"augmented_\" + str(i) + \"_\" + str(j) + '.wav', aug_sample, samplerate=sr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 117,
   "metadata": {},
   "outputs": [],
   "source": [
    "frame_length = int(0.05*sr)\r\n",
    "hop_length = int(0.04*sr)\r\n",
    "lastname_augs_features = []\r\n",
    "for y, fs in aug_samples:\r\n",
    "    mfcc = librosa.feature.mfcc(y, fs, n_mfcc=13, hop_length=hop_length, n_fft=frame_length, window=get_window('hamming',frame_length))\r\n",
    "    mfcc_delta = librosa.feature.delta(mfcc)\r\n",
    "    mfcc_delta2 = librosa.feature.delta(mfcc,order=2)\r\n",
    "    zcr = librosa.feature.zero_crossing_rate(y,frame_length=frame_length,hop_length=hop_length)\r\n",
    "    rms = librosa.feature.rms(y,frame_length=frame_length,hop_length=hop_length)\r\n",
    "    features = np.concatenate([mfcc,mfcc_delta,mfcc_delta2,zcr,rms])\r\n",
    "    lastname_augs_features.append(features)\r\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 118,
   "metadata": {},
   "outputs": [],
   "source": [
    "normalized_lastname_augs_features = cmvn(lastname_augs_features, True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div dir=\"rtl\">\r\n",
    "\r\n",
    "مدل مخفی مارکو را با داده های جدید حاصل از تکنیک افزونگی به علاوه داده اصلی آموزش میدهیم. سپس  مجددا میزان تطابق آن را با داده های تست خود میسنجیم.\r\n",
    "\r\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 161,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "GaussianHMM(covariance_type='full', n_components=5, n_iter=100)"
      ]
     },
     "execution_count": 161,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X = np.concatenate([np.transpose(item) for item in (\r\n",
    "    normalized_lastname_trains_features+lastname_augs_features)])\r\n",
    "n_components = 5\r\n",
    "lengths = np.array([item.shape[1]\r\n",
    "                    for item in normalized_lastname_trains_features])\r\n",
    "hmm_model = hmm.GaussianHMM(\r\n",
    "    n_components=n_components, covariance_type=\"full\", n_iter=100)\r\n",
    "hmm_model.fit(X, lengths)\r\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div dir=\"rtl\">\r\n",
    "\r\n",
    "برای نام خانوادگی تست (توسط گوینده یکسان با داده های آموزشی)\r\n",
    "\r\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 121,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-1962.4293289260004\n",
      "-2155.4162309995477\n",
      "-1553.391706658723\n"
     ]
    }
   ],
   "source": [
    "for item in normalized_lastname_tests_features:\r\n",
    "    print(hmm_model.score(np.transpose(item)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div dir=\"rtl\">\r\n",
    "\r\n",
    "برای نام ها\r\n",
    "\r\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 123,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-2373.971003380662\n",
      "-2445.445270637625\n",
      "-2270.2056591256874\n"
     ]
    }
   ],
   "source": [
    "for item in normalized_firstnames_features:\r\n",
    "    print(hmm_model.score(np.transpose(item)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div dir=\"rtl\">\r\n",
    "\r\n",
    "برای نام خانوادگی توسط گوینده ای دیگر\r\n",
    "\r\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 165,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-2095.3757821248487\n",
      "-1915.3024090532247\n",
      "-3633.244364656579\n"
     ]
    }
   ],
   "source": [
    "for item in normalized_lastname_another_features:\r\n",
    "    print(hmm_model.score(np.transpose(item)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div dir=\"rtl\">\r\n",
    "\r\n",
    "تحلیل داده های بالا به سادگی قابل انجام است. مشاهده میکنیم که برای داده های تست نام خانوادگی با گوینده یکسان کاملا اعداد بهبود پیدا کرده و با اعداد به دست آمده از نام قابل تمایز است. همچنین برای داده نام خانوادگی توسط گوینده دیگر نیز بهبود مشاهده می شود و قابل قبول است چون ما در یک هیچ یک از افزونگی هایی که ایجاد کردیم تغییر صدای گوینده را نداشتیم. با جمع آوری داده از افراد مختلف مطمئنا بهبود آن به شکل چشم گیری خواهد بود.\r\n",
    "\r\n",
    "## <center> با تشکر از توجه شما </center>\r\n",
    "</div>"
   ]
  }
 ],
 "metadata": {
  "interpreter": {
   "hash": "57baa5815c940fdaff4d14510622de9616cae602444507ba5d0b6727c008cbd6"
  },
  "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": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}