LoadAndPreprocessTest.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import LoadAndPreprocessDataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "sd_GSCmdV2/train/follow/14c7b073_nohash_1.wav.npy\n",
      "sd_GSCmdV2/train/seven/ae5578f4_nohash_0.wav.npy\n",
      "sd_GSCmdV2/train/up/29fb33da_nohash_1.wav.npy\n",
      "sd_GSCmdV2/train/five/72e382bd_nohash_0.wav.npy\n",
      "sd_GSCmdV2/train/backward/3f45a0cf_nohash_2.wav.npy\n",
      "sd_GSCmdV2/train/two/a1e71565_nohash_3.wav.npy\n",
      "sd_GSCmdV2/train/five/079d1020_nohash_0.wav.npy\n",
      "sd_GSCmdV2/train/tree/7f581e94_nohash_0.wav.npy\n",
      "sd_GSCmdV2/train/backward/1970b130_nohash_0.wav.npy\n",
      "sd_GSCmdV2/train/right/7d149b38_nohash_0.wav.npy\n",
      "sd_GSCmdV2/train/seven/5a9f7324_nohash_4.wav.npy\n",
      "sd_GSCmdV2/train/left/95218ce8_nohash_0.wav.npy\n",
      "sd_GSCmdV2/train/six/25132942_nohash_0.wav.npy\n",
      "sd_GSCmdV2/train/left/ddedba85_nohash_3.wav.npy\n",
      "sd_GSCmdV2/train/two/9ff2d2f4_nohash_3.wav.npy\n",
      "sd_GSCmdV2/train/seven/adb45ef2_nohash_0.wav.npy\n",
      "sd_GSCmdV2/train/left/24befdb3_nohash_0.wav.npy\n",
      "sd_GSCmdV2/train/four/eee5e541_nohash_0.wav.npy\n",
      "sd_GSCmdV2/train/four/be7a5b2d_nohash_0.wav.npy\n",
      "sd_GSCmdV2/train/left/de3fe344_nohash_0.wav.npy\n",
      "sd_GSCmdV2/train/bird/bc065a17_nohash_1.wav.npy\n",
      "sd_GSCmdV2/train/two/aff582a1_nohash_3.wav.npy\n",
      "sd_GSCmdV2/train/marvin/6f342826_nohash_0.wav.npy\n",
      "sd_GSCmdV2/train/forward/578d3efb_nohash_3.wav.npy\n",
      "sd_GSCmdV2/train/on/b76f6088_nohash_1.wav.npy\n",
      "sd_GSCmdV2/train/seven/72e382bd_nohash_3.wav.npy\n",
      "sd_GSCmdV2/train/up/f2f0d244_nohash_3.wav.npy\n",
      "sd_GSCmdV2/train/on/0e6e36c9_nohash_2.wav.npy\n",
      "sd_GSCmdV2/train/three/a24582a0_nohash_0.wav.npy\n",
      "sd_GSCmdV2/train/stop/9b8a7439_nohash_0.wav.npy\n",
      "sd_GSCmdV2/train/two/21e8c417_nohash_1.wav.npy\n",
      "sd_GSCmdV2/train/one/d3f22f0e_nohash_1.wav.npy\n",
      "sd_GSCmdV2/train/two/079d1020_nohash_3.wav.npy\n",
      "sd_GSCmdV2/train/five/b5cf6ea8_nohash_9.wav.npy\n",
      "sd_GSCmdV2/train/one/b4ea0d9a_nohash_6.wav.npy\n",
      "sd_GSCmdV2/train/three/9ff2d2f4_nohash_3.wav.npy\n",
      "sd_GSCmdV2/train/five/324210dd_nohash_4.wav.npy\n",
      "sd_GSCmdV2/train/wow/3a53b856_nohash_1.wav.npy\n",
      "sd_GSCmdV2/train/off/a8f45bdc_nohash_1.wav.npy\n",
      "sd_GSCmdV2/train/up/ee1d8d49_nohash_3.wav.npy\n",
      "sd_GSCmdV2/train/on/338dacf5_nohash_0.wav.npy\n",
      "sd_GSCmdV2/train/nine/cae62f38_nohash_0.wav.npy\n",
      "sd_GSCmdV2/train/up/74241b28_nohash_3.wav.npy\n",
      "sd_GSCmdV2/train/yes/78030270_nohash_0.wav.npy\n",
      "sd_GSCmdV2/train/house/ba676390_nohash_2.wav.npy\n",
      "sd_GSCmdV2/train/backward/42a99aec_nohash_3.wav.npy\n",
      "sd_GSCmdV2/train/two/821b64cc_nohash_1.wav.npy\n",
      "sd_GSCmdV2/train/five/3d9bbe2d_nohash_2.wav.npy\n",
      "sd_GSCmdV2/train/go/5a0bc987_nohash_0.wav.npy\n",
      "sd_GSCmdV2/train/left/3b195250_nohash_3.wav.npy\n",
      "Loading  100  files (seconds): 1.2359576225280762\n",
      "Size of X:\t12800000 bytes\n",
      "Shape of X:\t (100, 16000)\n",
      "Transforming  100  elements (seconds): 0.3217647075653076\n",
      "Size of trasformed: \t8064000 bytes\n",
      "Shape of transformed:\t (100, 80, 126)\n",
      "(100, 80, 126)\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "20\n",
      "17\n",
      "2\n",
      "15\n",
      "20\n",
      "12\n",
      "15\n",
      "20\n",
      "20\n",
      "5\n",
      "17\n",
      "4\n",
      "16\n",
      "4\n",
      "12\n",
      "17\n",
      "4\n",
      "14\n",
      "14\n",
      "4\n",
      "20\n",
      "12\n",
      "20\n",
      "20\n",
      "6\n",
      "17\n",
      "2\n",
      "6\n",
      "13\n",
      "8\n",
      "12\n",
      "11\n",
      "12\n",
      "15\n",
      "11\n",
      "13\n",
      "15\n",
      "20\n",
      "7\n",
      "2\n",
      "6\n",
      "19\n",
      "2\n",
      "0\n",
      "20\n",
      "20\n",
      "12\n",
      "15\n",
      "9\n",
      "4\n"
     ]
    }
   ],
   "source": [
    "import time\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "#Load filenames\n",
    "train,val,test = LoadAndPreprocessDataset.loadDatasetFilenames(nCategories=21)\n",
    "\n",
    "#batch_size=len(test)\n",
    "batch_size=100\n",
    "\n",
    "start = time.time()\n",
    "#Load Trainfiles\n",
    "X,y=LoadAndPreprocessDataset.loadBatch(train,batch_size=batch_size,nCategories=21)\n",
    "end = time.time()\n",
    "print(\"Loading \", batch_size, \" files (seconds):\", end - start)\n",
    "print(\"Size of X:\\t%d bytes\" % (X.nbytes))\n",
    "print(\"Shape of X:\\t\",X.shape)\n",
    "\n",
    "start = time.time()\n",
    "#Extract features\n",
    "#features=LoadAndPreprocessDataset.MFCC_DELTA_12(X)\n",
    "features=LoadAndPreprocessDataset.melspect(X)\n",
    "end = time.time()\n",
    "print(\"Transforming \", batch_size, \" elements (seconds):\", end - start)\n",
    "print(\"Size of trasformed: \\t%d bytes\" % (features.nbytes))\n",
    "print(\"Shape of transformed:\\t\",features.shape)\n",
    "\n",
    "print(features.shape)\n",
    "plt.figure(figsize=(10, 4))\n",
    "plt.pcolormesh(features[0])\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "del(X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.figure(figsize=(10, 4))\n",
    "plt.pcolormesh(features[10])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import librosa # mfcc functions\n",
    "\n",
    "sr=16000\t#Sampling rate of audiofiles\n",
    "\n",
    "\n",
    "y=np.load(\"sd_GSCmdV2/test/right/3d86b69a_nohash_2.wav.npy\")\n",
    "\n",
    "print(np.max(y))\n",
    "print(np.mean(np.abs(y)))\n",
    "\n",
    "import sounddevice as sd\n",
    "sd.play(y, sr)\n",
    "\n",
    "S = librosa.feature.melspectrogram(y, sr=sr, n_fft=1024,\n",
    "\t\t\t\t\t\t\t\t\t\thop_length=128, power=1.0,\n",
    "\t\t\t\t\t\t\t\t\t\tn_mels=80, fmin=40.0, fmax=sr/2)\n",
    "\n",
    "# Convert to log scale (dB). We'll use the peak power (max) as reference.\n",
    "log_S = librosa.power_to_db(S, ref=np.max)\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "plt.figure(figsize=(10,4))\n",
    "plt.plot(y)\n",
    "plt.xlabel(\"Samples\")\n",
    "plt.ylabel(\"Intensity\")\n",
    "plt.xlim(0, len(y))\n",
    "plt.show()\n",
    "\n",
    "plt.figure(figsize=(10, 4))\n",
    "plt.pcolormesh(log_S)\n",
    "plt.show()\n",
    "\n",
    "mfcc        = librosa.feature.mfcc(S=log_S, n_mfcc=40)\n",
    "plt.figure(figsize=(10, 4))\n",
    "plt.pcolormesh(mfcc)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "gputest",
   "language": "python",
   "name": "gputest"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}