Join Audio.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "import sys\n",
    "import os\n",
    "import glob\n",
    "import librosa\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from tqdm import tqdm\n",
    "\n",
    "from matplotlib import pyplot as plt\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "INPUT_FOLDER = '../ryan_audio/les_mis/'\n",
    "OUTPUT_NAME = 'les_mis.mp3'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# Get all audio files into single array\n",
    "for i,f in enumerate(glob.glob(INPUT_FOLDER+'*')):\n",
    "    temp, sr = librosa.load(f,sr=None)\n",
    "    if i == 0:\n",
    "        final = temp\n",
    "    else:\n",
    "        final = np.concatenate((final,temp))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Save array\n",
    "librosa.output.write_wav(OUTPUT_NAME,final,sr)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Open Audacity and split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "LANGUAGE_CODE = 'fr'\n",
    "BOOK_TITLE = 'lesmis'\n",
    "LABELS_NAME = 'lesmis_labels.txt'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "256"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Make language directory\n",
    "os.system('mkdir '+LANGUAGE_CODE)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Make book directory\n",
    "os.system('mkdir '+LANGUAGE_CODE+'/'+BOOK_TITLE)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# Read silences\n",
    "silences = np.array(pd.read_csv(LABELS_NAME,delimiter='\\t',header=None).iloc[:,0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "11099it [00:11, 950.90it/s]\n"
     ]
    }
   ],
   "source": [
    "# Split data and write\n",
    "for i,s in tqdm(enumerate(silences)):\n",
    "#     if i < 10:\n",
    "    if i == 0:\n",
    "        prev_cursor = 0\n",
    "    curr_cursor = int(sr*s)\n",
    "    clip = final[prev_cursor:curr_cursor]\n",
    "    output_num = str(i).zfill(4)\n",
    "    librosa.output.write_wav(LANGUAGE_CODE+'/'+BOOK_TITLE+'/'+BOOK_TITLE+output_num,clip,sr)\n",
    "    prev_cursor = curr_cursor\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "11099it [00:00, 476168.16it/s]\n"
     ]
    }
   ],
   "source": [
    "# Get statistics\n",
    "durations = []\n",
    "for i,s in tqdm(enumerate(silences)):\n",
    "#     if i < 10:\n",
    "    if i == 0:\n",
    "        prev_cursor = 0\n",
    "    curr_cursor = s\n",
    "    durations.append(curr_cursor-prev_cursor)\n",
    "    prev_cursor = curr_cursor"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAhYAAAFkCAYAAAB8RXKEAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAAPYQAAD2EBqD+naQAAIABJREFUeJzt3X+cZXV93/HXG3F3C+ku2pVdqd1HpCSb6aMtcYeCNIHG\nbBpiMMSUtmHIlEehqVV+hE5qg7baUEnTBCtLEEyp0qoBJw+6xGKUsFFiqUFkK4Oa6LgtDTgi7JoJ\ny+wG3EV2v/3jnIl3rzOzzOz3ztzZfT0fj/tg5ns+c+73Hs7OvO/3fr/npJSCJElSDcctdQckSdLR\nw2AhSZKqMVhIkqRqDBaSJKkag4UkSarGYCFJkqoxWEiSpGoMFpIkqRqDhSRJqsZgIUmSqpl3sEhy\nTpKPJflGkoNJLujafmKSm5N8PclzSb6c5F901axMckuSySR7k2xNcnJXzcuS3JFkKsnuJB9IcuLC\nXqYkSVoMCxmxOBH4AnA5MNONRrYAPw5cDPxA+/3NSd7QUXMjcD5wIXAucApwV9d+PgIMAJvb2nOB\nWxfQX0mStEhyJDchS3IQeGMp5WMdbX8E/HYp5T90tH0euKeU8u+SrAb+FLiolPLRdvtGYBx4bSll\ne5IB4MvAYCnlkbbmPOATwKtKKTsX3GlJktQzvZhj8VnggiSnACR5HfB9wLZ2+yBwPHDf9A+UUnYA\nE8DZbdNrgd3ToaL1KZoRkrN60GdJklTB8T3Y51XAfwGeSPICcAD456WUB9rt64HnSyl7un5uV7tt\nuuabnRtLKQeSPN1Rc4gkfwU4D3gc2FfhdUiSdKxYBXwvsK2U8mdHsqNeBItfoBlVeAPNKMS5wPuS\nPFlK+YMePN+084A7erh/SZKOdj9HM8dxwaoGiySrgP9AM+/i99rmP07yGuCtwB8AO4EVSVZ3jVqs\na7fR/rd7lchLgJd31HR7HOD2229nYGCgwqs5doyMjLBly5al7say4jFbGI/b/HnMFsbjNj/j4+MM\nDw9D+7f0SNQesXhp+zjQ1X6A78zneBh4gWa1R+fkzQ3Ag23Ng8BJSV7TMc9iMxDgoVmeex/AwMAA\nmzZtOvJXcgxZs2aNx2yePGYL43GbP4/ZwnjcFuyIpxLMO1i015I4jeaPPMCpSU4Hni6lfD3J/cB/\nSnIV8DXgR4BLgH8JUErZk+Q24IYku4G9wE3AA6WU7W3NV5NsA96f5C3ACuC9wKgrQiRJ6l8LGbE4\nA/g0zQqNArynbf8QcBnws8B/BG6n+ejia8DbSyn/pWMfIzSjGFuBlcC9wBVdz3MxcDPNapCDbe3V\nC+ivJElaJPMOFqWU+5ljmWop5ZvAPzvMPvbTrB65ao6aZ4Dh+fZPkiQtHe8VIoaGhpa6C8uOx2xh\nPG7z5zFbGI/b0jmiK2/2kySbgIcffvhhJ+xIkjQPY2NjDA4OQnPF67Ej2ZcjFpIkqRqDhSRJqsZg\nIUmSqjFYSJKkagwWkiSpGoOFJEmqxmAhSZKqMVhIkqRqDBaSJKkag4UkSarGYCFJkqoxWEiSpGoM\nFpIkqRqDhSRJqsZgIUmSqjFYSJKkagwWkiSpGoOFJEmq5vil7sDRamJigsnJyTlr1q5dy4YNGxap\nR5Ik9Z7BogcmJibYuHGAffuem7Nu1aoT2LFj3HAhSTpqGCx6YHJysg0VtwMDs1SNs2/fMJOTkwYL\nSdJRY95zLJKck+RjSb6R5GCSC2aoGUhyd5Jnkvx5koeSvKpj+8oktySZTLI3ydYkJ3ft42VJ7kgy\nlWR3kg8kOXFhL3OpDACbZnnMFjgkSVq+FjJ580TgC8DlQOnemOSvA58BvgKcC/wt4DpgX0fZjcD5\nwIVtzSnAXV27+gjNX9/Nbe25wK0L6K8kSVok8/4opJRyL3AvQJLMUPIrwCdKKW/vaHts+oskq4HL\ngItKKfe3bZcC40nOLKVsTzIAnAcMllIeaWuuAj6R5K2llJ3z7bckSeq9qstN26BxPvB/k9ybZFeS\nzyX56Y6yQZpAc990QyllBzABnN02vRbYPR0qWp+iGSE5q2afJUlSPbWvY3Ey8D3ANcA9wN8HPgr8\nTpJz2pr1wPOllD1dP7ur3TZd883OjaWUA8DTHTWSJKnP1F4VMh1U/kcp5ab26y8l+bvAm2nmXkiS\npKNU7WAxCbwAjHe1jwM/1H69E1iRZHXXqMW6dtt0TfcqkZcAL++omdHIyAhr1qw5pG1oaIihoaF5\nvAxJko5Oo6OjjI6OHtI2NTVVbf9Vg0Up5dtJ/jewsWvT9wNfa79+mCZ8bKb5mIQkG4ENwINtzYPA\nSUle0zHPYjMQ4KG5+rBlyxY2bdp0pC9FkqSj0kxvtsfGxhgcHKyy/3kHi/ZaEqfR/JEHODXJ6cDT\npZSvA+8GfjvJZ4BPA68H3gD8PYBSyp4ktwE3JNkN7AVuAh4opWxva76aZBvw/iRvAVYA7wVGXREi\nSVL/WsiIxRk0gaG0j/e07R8CLiul/I8kbwb+DfAbwA7gH5RSHuzYxwhwANgKrKRZvnpF1/NcDNxM\nsxrkYFt79QL6K0mSFslCrmNxP4dZTVJK+SDwwTm27weuah+z1TwDDM+3f5Ikael4r5AFONydS8fH\nu+euSpJ0bDBYzNOLvXOpJEnHIoPFPL24O5feA7xz8TolSVKfMFgs2PSdS2fiRyGSpGNT7Ut6S5Kk\nY5jBQpIkVWOwkCRJ1RgsJElSNQYLSZJUjcFCkiRVY7CQJEnVGCwkSVI1BgtJklSNwUKSJFVjsJAk\nSdUYLCRJUjUGC0mSVI3BQpIkVWOwkCRJ1RgsJElSNQYLSZJUjcFCkiRVc/xSd+BYNz4+Puf2tWvX\nsmHDhkXqjSRJR8ZgsWSeAo5jeHh4zqpVq05gx45xw4UkaVmY90chSc5J8rEk30hyMMkFc9T+57bm\nF7raVya5Jclkkr1JtiY5uavmZUnuSDKVZHeSDyQ5cb797V/PAAeB24GHZ3nczr59zzE5OblkvZQk\naT4WMmJxIvAF4Dbgd2YrSvIzwFnAN2bYfCPweuBCYA9wC3AXcE5HzUeAdcBmYAXwQeBWYO63+MvO\nALBpqTshSVIV8w4WpZR7gXsBkmSmmiR/FfgN4Dzgnq5tq4HLgItKKfe3bZcC40nOLKVsTzLQ/uxg\nKeWRtuYq4BNJ3lpK2TnffkuSpN6rviqkDRsfBq4vpcw0M3GQJtDcN91QStkBTABnt02vBXZPh4rW\np4BCMwoiSZL6UC+Wm74NeL6UcvMs29e32/d0te9qt03XfLNzYynlAPB0R40kSeozVVeFJBkEfgF4\nTc39zsfIyAhr1qw5pG1oaIihoaEl6pEkSf1jdHSU0dHRQ9qmpqaq7b/2ctMfBl4BfL1j+sVLgBuS\n/MtSyqnATmBFktVdoxbr2m20/+1eJfIS4OUdNTPasmULmzY5GVKSpJnM9GZ7bGyMwcHBKvuv/VHI\nh4G/DZze8XgSuJ5mMiY06yhfoFntAUCSjcAG4MG26UHgpCSdIx+bgQAPVe6zJEmqZN4jFu21JE6j\n+SMPcGqS04GnSylfB3Z31X8b2FlK+b8ApZQ9SW6jGcXYDewFbgIeKKVsb2u+mmQb8P4kb6FZbvpe\nYNQVIZIk9a+FfBRyBvBpmhUaBXhP2/4hmmWk3coMbSPAAWArsJJm+eoVXTUXAzfTrAY52NZevYD+\nSpKkRbKQ61jczzw+QmnnVXS37Qeuah+z/dwzHHUXw5Ik6ejm3U0lSVI1BgtJklSNwUKSJFVjsJAk\nSdUYLCRJUjUGC0mSVI3BQpIkVWOwkCRJ1RgsJElSNQYLSZJUjcFCkiRVY7CQJEnVGCwkSVI1BgtJ\nklSNwUKSJFVjsJAkSdUYLCRJUjUGC0mSVI3BQpIkVWOwkCRJ1RgsJElSNQYLSZJUjcFCkiRVM+9g\nkeScJB9L8o0kB5Nc0LHt+CS/nuRLSf68rflQkld27WNlkluSTCbZm2RrkpO7al6W5I4kU0l2J/lA\nkhMX/lIlSVKvLWTE4kTgC8DlQOnadgLwg8C/B14D/AywEbi7q+5G4HzgQuBc4BTgrq6ajwADwOa2\n9lzg1gX0V5IkLZLj5/sDpZR7gXsBkqRr2x7gvM62JFcCDyV5VSnliSSrgcuAi0op97c1lwLjSc4s\npWxPMtDuZ7CU8khbcxXwiSRvLaXsnPcrlSRJPbcYcyxOohnZeKb9fpAm0Nw3XVBK2QFMAGe3Ta8F\ndk+Hitan2v2c1esOS5KkhelpsEiyEvg14COllD9vm9cDz7ejG512tduma77ZubGUcgB4uqNGkiT1\nmZ4FiyTHA/+dZpTh8l49jyRJ6h/znmPxYnSEir8G/GjHaAXATmBFktVdoxbr2m3TNd2rRF4CvLyj\nZkYjIyOsWbPmkLahoSGGhoYW8lIkSTqqjI6OMjo6ekjb1NRUtf1XDxYdoeJU4HWllN1dJQ8DL9Cs\n9vho+zMbgQ3Ag23Ng8BJSV7TMc9iMxDgobmef8uWLWzatKnGS5Ek6agz05vtsbExBgcHq+x/3sGi\nvZbEaTR/5AFOTXI6zfyHp2iWjf4g8AbgpUnWtXVPl1K+XUrZk+Q24IYku4G9wE3AA6WU7QCllK8m\n2Qa8P8lbgBXAe4FRV4RIktS/FjJicQbwaZq5EwV4T9v+IZrrV/xU2/6Ftj3t968D/lfbNgIcALYC\nK2mWr17R9TwXAzfTrAY52NZevYD+SpKkRbKQ61jcz9yTPg87IbSUsh+4qn3MVvMMMDzf/kmSpKXj\nvUIkSVI1BgtJklSNwUKSJFVjsJAkSdUYLCRJUjUGC0mSVI3BQpIkVWOwkCRJ1RgsJElSNT25u6nq\nGh8fn3P72rVr2bBhwyL1RpKk2Rks+tpTwHEMD899ZfNVq05gx45xw4UkackZLPraMzT3X7sdGJil\nZpx9+4aZnJw0WEiSlpzBYlkYADYtdSckSTosJ29KkqRqDBaSJKkag4UkSarGYCFJkqoxWEiSpGoM\nFpIkqRqDhSRJqsZgIUmSqjFYSJKkagwWkiSpGoOFJEmqZt7BIsk5ST6W5BtJDia5YIaadyV5Mslz\nST6Z5LSu7SuT3JJkMsneJFuTnNxV87IkdySZSrI7yQeSnDj/lyhJkhbLQkYsTgS+AFwOlO6NSa4B\nrgTeBJwJPAtsS7Kio+xG4HzgQuBc4BTgrq5dfYTm7lub29pzgVsX0F9JkrRI5n1301LKvcC9AEky\nQ8nVwHWllI+3NZcAu4A3AncmWQ1cBlxUSrm/rbkUGE9yZille5IB4DxgsJTySFtzFfCJJG8tpeyc\nb78lSVLvVZ1jkeTVwHrgvum2Usoe4CHg7LbpDJpA01mzA5joqHktsHs6VLQ+RTNCclbNPkuSpHpq\nT95cT/PHf1dX+652G8A64Pk2cMxWsx74ZufGUsoB4OmOGkmS1Gfm/VFIvxsZGWHNmjWHtA0NDTE0\nNLREPZIkqX+Mjo4yOjp6SNvU1FS1/dcOFjuB0IxKdI5arAMe6ahZkWR116jFunbbdE33KpGXAC/v\nqJnRli1b2LRp04JfgCRJR7OZ3myPjY0xODhYZf9VPwoppTxG84d/83RbO1nzLOCzbdPDwAtdNRuB\nDcCDbdODwElJXtOx+800oeWhmn2WJEn1zHvEor2WxGk0f+QBTk1yOvB0KeXrNEtJ35HkUeBx4Drg\nCeBuaCZzJrkNuCHJbmAvcBPwQClle1vz1STbgPcneQuwAngvMOqKEEmS+tdCPgo5A/g0zSTNAryn\nbf8QcFkp5fokJ9Bcc+Ik4DPA60spz3fsYwQ4AGwFVtIsX72i63kuBm6mWQ1ysK29egH9lSRJi2Qh\n17G4n8N8hFJKuRa4do7t+4Gr2sdsNc8Aw/PtnyRJWjreK0SSJFVjsJAkSdUYLCRJUjUGC0mSVI3B\nQpIkVWOwkCRJ1RgsJElSNUfdTciO1MTEBJOTk7NuHx8fX8TeSJK0vBgsOkxMTLBx4wD79j231F2Z\nt8MFnrVr17Jhw4ZF6o0k6VhlsOgwOTnZhorbgYFZqu4B3rl4nTqsp4DjGB6e+yKlq1adwI4d44YL\nSVJPGSxmNADMduv1fvso5BmaW6nMFYbG2bdvmMnJSYOFJKmnDBZHjbnCkCRJi8NVIZIkqRqDhSRJ\nqsZgIUmSqjFYSJKkagwWkiSpGoOFJEmqxmAhSZKqMVhIkqRqDBaSJKkag4UkSarGYCFJkqqpHiyS\nHJfkuiR/kuS5JI8meccMde9K8mRb88kkp3VtX5nkliSTSfYm2Zrk5Nr9lSRJ9fRixOJtwL8ALgd+\nAPgl4JeSXDldkOQa4ErgTcCZwLPAtiQrOvZzI3A+cCFwLnAKcFcP+itJkirpxd1NzwbuLqXc234/\nkeRimgAx7WrgulLKxwGSXALsAt4I3JlkNXAZcFEp5f625lJgPMmZpZTtPei3JEk6Qr0YsfgssDnJ\n9wEkOR34IeCe9vtXA+uB+6Z/oJSyB3iIJpQAnEETejprdgATHTWSJKnP9GLE4teA1cBXkxygCS//\ntpTy2+329UChGaHotKvdBrAOeL4NHLPVSJKkPtOLYPGzwMXARcBXgB8EfiPJk6WU3+rB80mSpD7R\ni2BxPfAfSyn/vf3+y0m+F3g78FvATiA0oxKdoxbrgEfar3cCK5Ks7hq1WNdum9XIyAhr1qw5pG1o\naIihoaEFvRhJko4mo6OjjI6OHtI2NTVVbf+9CBYnAAe62g7SzucopTyWZCewGfgSQDtZ8yzglrb+\nYeCFtuajbc1GYAPw4FxPvmXLFjZt2lTlhUiSdLSZ6c322NgYg4ODVfbfi2Dxu8A7kjwBfBnYBIwA\nH+ioubGteRR4HLgOeAK4G5rJnEluA25IshvYC9wEPOCKEEmS+lcvgsWVNEHhFuBk4EngN9s2AEop\n1yc5AbgVOAn4DPD6UsrzHfsZoRn52AqsBO4FruhBfyVJUiXVg0Up5VngF9vHXHXXAtfOsX0/cFX7\nkCRJy4D3CpEkSdUYLCRJUjUGC0mSVI3BQpIkVWOwkCRJ1RgsJElSNb24joX61Pj4+GFr1q5dy4YN\nGxahN5Kko5HB4pjwFHAcw8PDh61cteoEduwYN1xIkhbEYHFMeIbmdi23AwNz1I2zb98wk5OTBgtJ\n0oIYLI4pAzS3bpEkqTecvClJkqoxWEiSpGoMFpIkqRqDhSRJqsZgIUmSqjFYSJKkagwWkiSpGoOF\nJEmqxmAhSZKqMVhIkqRqDBaSJKkag4UkSarGYCFJkqoxWEiSpGp6EiySnJLkt5JMJnkuyReTbOqq\neVeSJ9vtn0xyWtf2lUluafexN8nWJCf3or+SJKmO6sEiyUnAA8B+4DxgAPhXwO6OmmuAK4E3AWcC\nzwLbkqzo2NWNwPnAhcC5wCnAXbX7K0mS6jm+B/t8GzBRSvn5jravddVcDVxXSvk4QJJLgF3AG4E7\nk6wGLgMuKqXc39ZcCownObOUsr0H/ZYkSUeoFx+F/BTw+SR3JtmVZCzJX4SMJK8G1gP3TbeVUvYA\nDwFnt01n0ISezpodwERHjSRJ6jO9CBanAm8BdgA/DvwmcFOSf9JuXw8UmhGKTrvabQDrgOfbwDFb\njSRJ6jO9+CjkOGB7KeWd7fdfTPI3gTcDv9WD5zvEyMgIa9asOaRtaGiIoaGhXj+1JEl9b3R0lNHR\n0UPapqamqu2/F8HiKWC8q20c+Aft1zuB0IxKdI5arAMe6ahZkWR116jFunbbrLZs2cKmTZvmKpEk\n6Zg105vtsbExBgcHq+y/F8HiAWBjV9tG2gmcpZTHkuwENgNfAmgna54F3NLWPwy80NZ8tK3ZCGwA\nHuxBn9VhfLw7Fx5q7dq1bNiwYZF6I0laTnoRLLYADyR5O3AnTWD4eeCfd9TcCLwjyaPA48B1wBPA\n3dBM5kxyG3BDkt3AXuAm4AFXhPTSU8BxDA8Pz1m1atUJ7NgxbriQJH2X6sGilPL5JD8D/BrwTuAx\n4OpSym931Fyf5ATgVuAk4DPA60spz3fsagQ4AGwFVgL3AlfU7q86PQMcBG6nufzITMbZt2+YyclJ\ng4Uk6bv0YsSCUso9wD2HqbkWuHaO7fuBq9qHFtUA4DwVSdL8ea8QSZJUjcFCkiRVY7CQJEnVGCwk\nSVI1BgtJklSNwUKSJFVjsJAkSdUYLCRJUjUGC0mSVI3BQpIkVWOwkCRJ1RgsJElSNQYLSZJUjcFC\nkiRVY7CQJEnVGCwkSVI1BgtJklSNwUKSJFVjsJAkSdUYLCRJUjUGC0mSVI3BQpIkVWOwkCRJ1Rzf\n6ydI8jbgV4EbSym/2NH+LuDngZOAB4C3lFIe7di+ErgB+FlgJbANuLyU8s1e91mHNz4+Puf2tWvX\nsmHDhkXqjSSpX/Q0WCT5O8CbgC92tV8DXAlcAjwO/AqwLclAKeX5tuxG4PXAhcAe4BbgLuCcXvZZ\nh/MUcBzDw8NzVq1adQI7dowbLiTpGNOzYJHke4DbaUYl3tm1+WrgulLKx9vaS4BdwBuBO5OsBi4D\nLiql3N/WXAqMJzmzlLK9V/3W4TwDHKT5XzswS804+/YNMzk5abCQpGNML0csbgF+t5TyB0n+Ilgk\neTWwHrhvuq2UsifJQ8DZwJ3AGW3fOmt2JJloawwWS24A2LTUnZAk9ZmeBIskFwE/SBMQuq0HCs0I\nRadd7TaAdcDzpZQ9c9RIkqQ+Uz1YJHkVzfyIHyulfLv2/iVJUv/qxYjFIPAKYCxJ2raXAOcmuRL4\nASA0oxKdoxbrgEfar3cCK5Ks7hq1WNdum9XIyAhr1qw5pG1oaIihoaEFvhwtlCtHJKn/jI6OMjo6\nekjb1NRUtf33Ilh8CvhbXW0fBMaBXyul/EmSncBm4EsA7WTNs2jmZQA8DLzQ1ny0rdkIbAAenOvJ\nt2zZwqZNfva/tFw5Ikn9aqY322NjYwwODlbZf/VgUUp5FvhKZ1uSZ4E/K6VMv4W9EXhHkkdplpte\nBzwB3N3uY0+S24AbkuwG9gI3AQ+4ImQ5cOWIJB2ren6BrFY55JtSrk9yAnArzQWyPgO8vuMaFgAj\nwAFgK80Fsu4Frlic7qoOV45I0rFmUYJFKeVHZ2i7Frh2jp/ZD1zVPiRJ0jLgvUIkSVI1BgtJklSN\nwUKSJFVjsJAkSdUYLCRJUjUGC0mSVI3BQpIkVWOwkCRJ1RgsJElSNQYLSZJUjcFCkiRVY7CQJEnV\nGCwkSVI1BgtJklSNwUKSJFVjsJAkSdUcv9Qd0LFtfHx8zu1r165lw4YNi9QbSdKRMlhoiTwFHMfw\n8PCcVatWncCOHeOGC0laJgwWWiLPAAeB24GBWWrG2bdvmMnJSYOFJC0TBgstsQFg01J3QpJUiZM3\nJUlSNQYLSZJUjcFCkiRVY7CQJEnVVA8WSd6eZHuSPUl2Jfloku+foe5dSZ5M8lySTyY5rWv7yiS3\nJJlMsjfJ1iQn1+6vJEmqpxcjFucA7wXOAn4MeCnw+0n+0nRBkmuAK4E3AWcCzwLbkqzo2M+NwPnA\nhcC5wCnAXT3oryRJqqT6ctNSyk92fp/knwLfBAaBP2ybrwauK6V8vK25BNgFvBG4M8lq4DLgolLK\n/W3NpcB4kjNLKdtr91uSJB25xbiOxUlAAZ4GSPJqYD1w33RBKWVPkoeAs4E7gTPavnXW7Egy0dYY\nLI4hXvZbkpaPngaLJKH5SOMPSylfaZvX0wSNXV3lu9ptAOuA50spe+ao0VHPy35L0nLT6xGL9wF/\nA/ihHj/PXxgZGWHNmjWHtA0NDTE0NLRYXVA1XvZbkmobHR1ldHT0kLapqalq++9ZsEhyM/CTwDml\nlKc6Nu0EQjMq0TlqsQ54pKNmRZLVXaMW69pts9qyZQubNnmJ6KOLl/2WpFpmerM9NjbG4OBglf33\n5DoWbaj4aeB1pZSJzm2llMdowsHmjvrVNKtIPts2PQy80FWzEdgAPNiLPkuSpCNXfcQiyfuAIeAC\n4Nkk69pNU6WUfe3XNwLvSPIo8DhwHfAEcDf8xWTO24AbkuwG9gI3AQ+4IkSSpP7Vi49C3kwzOfN/\ndrVfCnwYoJRyfZITgFtpVo18Bnh9KeX5jvoR4ACwFVgJ3AtcsdBO7dy5k5/4iZ/iT/90ctaa55/f\nv9DdS5IkenMdixf18Uop5Vrg2jm27weuah9HbGxsjC9+8fPALwB/eZaqz9OsRJAkSQuxGNex6DPX\n0FzEcya3ANsWsS+SJB1djsFgoaORF9GSpP5gsNAy50W0JKmfGCy0zHkRLUnqJwYLHSW8iJYk9YOe\nXCBLkiQdmxyx0DHDCZ6S1HsGCx0DnOApSYvFYKFjgBM8JWmxGCx0DHGCpyT1mpM3JUlSNQYLSZJU\njcFCkiRV4xwLqYNLUiXpyBgsJODFLklduXIVd921lVe+8pWz1hg+JB3LDBYS8OKWpH6G/ft/kTe8\n4Q1z7snrYUg6lhkspEPMtSR1HK+HIUlzM1hI8+b1MCRpNgYLqQcONwl0//79rFy5cs4a52pIWo4M\nFlJVL24SKLwEODBnhXM1JC1HBgupqhczCfQe4J2HqXGuhqTlyWAh9cThJoEerkaSlieDhYBRYGip\nO7HM9Mcxm5iYYHJycs6afpqrMTo6ytDQ0h+35cRjtjAet6XT98EiyRXAW4H1wBeBq0op/3tpe3W0\n6Y8/ksvL4hyzuSaBPvXUU1x44T9i//5vzbmPfrqol7/s589jtjAet6XT18Eiyc8C7wHeBGwHRoBt\nSb6/lDL32zRpWXuxk0ChxkW9aoWPw42gTE1NMTEx0TcjKJLq6+tgQRMkbi2lfBggyZuB84HLgOuX\nsmNSb81nEuiRXtSrTvh4sSMoGzcOuNpFOor1bbBI8lJgEPjV6bZSSknyKeDsJeuYtKhezCTQGvup\nEz4ac+3n59m375HDrnZ5MXNHvBaI1J/6NlgAa2kW++/qat8FbJyhfhXM/pn0o48+2n51O3DSLE/5\nh+1/72H2X9oPHIU1TwB39Fmf+r1m+pj1S39q1Dw2y3aAHTTh458Bs31c8kfA3YfZz3NNb+65Z9Z/\nq5OTk/zrf/02vv3tfXPsB+C4tk+ze+lLV/Lud/86a9eunX0vxx3HwYNz72cpa5544gnuuOOOOWv6\nrc/9UDPkyyS/AAAFJklEQVTTcVvK/vRTzdq1a3nFK15xSFvHv8dVc/7wi5BSypHuoyeSvBL4BnB2\nKeWhjvZfB84tpZzdVX8xM/+mlyRJL87PlVI+ciQ76OcRi0maSxOu62pfB+ycoX4b8HPA48Dh3upI\nkqTvWAV8L83f0iPStyMWAEk+BzxUSrm6/T7ABHBTKeXdS9o5SZL0Xfp5xALgBuCDSR7mO8tNTwA+\nuJSdkiRJM+vrYFFKuTPJWuBdNB+BfAE4r5Typ0vbM0mSNJO+/ihEkiQtL8ctdQckSdLRw2AhSZKq\nOSqCRZIrkjyW5FtJPpfk7yx1n/pZkl9OcrDr8ZWl7le/SXJOko8l+UZ7jC6YoeZdSZ5M8lySTyY5\nbSn62i8Od8yS/LcZzr17lqq//SDJ25NsT7Inya4kH03y/TPUea51eDHHzfPtUEnenOSLSabax2eT\n/ERXzRGfZ8s+WHTcqOyXgdfQ3AF1WzvpU7P7Y5oJsevbxw8vbXf60ok0E4YvB75rMlKSa4AraW6S\ndybwLM25t2IxO9ln5jxmrd/j0HPvWL8F5TnAe4GzgB8DXgr8fpK/NF3guTajwx63lufbd3wduIbm\n+v6DwB8AdycZgIrnWSllWT+AzwG/0fF9aK63/EtL3bd+fdCEsLGl7sdyetBcO/qCrrYngZGO71cD\n3wL+8VL3tx8esxyz/wb8zlL3rZ8fNLczOAj8cEeb59rCjpvn2+GP258Bl7ZfVznPlvWIRceNyu6b\nbivN0fBGZYf3fe1w9f9LcnuSv7bUHVpOkrya5t1P57m3B3gIz73D+ZF26PqrSd6X5OVL3aE+cxLN\naM/T4Lk2D4cctw6ebzNIclySi2iuDfXZmufZsg4WzH2jsvWL351l43PAPwXOA94MvBr4X0lOXMpO\nLTPraX6Jee7Nz+8BlwA/CvwS8PeAe9qr6h7z2uNwI/CHpZTpeU+ea4cxy3EDz7fvkuRvJtkL7Afe\nB/xMKWUHFc+zvr5AlnqjlNJ5Lfg/TrId+Brwj2mGDqWeKKXc2fHtl5P8EfD/gB8BPr0kneov7wP+\nBvBDS92RZWbG4+b5NqOvAqcDa4B/CHw4ybk1n2C5j1jM90ZlmkEpZQr4P8AxPct8nnbSzOfx3DsC\npZTHaP4dH/PnXpKbgZ8EfqSU8lTHJs+1Ocxx3L6L5xuUUl4opfxJKeWRUsq/pVnwcDUVz7NlHSxK\nKd8GHgY2T7e1Q1ybgc8uVb+WmyTfQ/MPbc5/lPqO9hfUTg4991bTzFD33HuRkrwK+Csc4+de+8fx\np4HXlVImOrd5rs1uruM2S73n23c7DlhZ8zw7Gj4K8UZl85Tk3cDv0nz88VeBfw98Gxhdyn71m3bO\nyWk0KR7g1CSnA0+XUr5O85nuO5I8CjwOXEezIunuJehuX5jrmLWPXwbuovkFdhrw6zSjZUd8q+bl\nKsn7aJZAXgA8m2T6HeNUKWVf+7XnWpfDHbf2XPR865DkV2nmnUwAfxn4OZp5Jz/eltQ5z5Z6qUul\n5TKXtwfhW8CDwBlL3ad+ftAEiCfa4zUBfAR49VL3q98e7T+4gzQft3U+/mtHzbU0S7Seo/llddpS\n97tfjxmwCriX5pf8PuBPgN8EXrHU/V7iYzbT8ToAXNJV57k2j+Pm+TbjMftAexy+1R6X3wd+tKvm\niM8zb0ImSZKqWdZzLCRJUn8xWEiSpGoMFpIkqRqDhSRJqsZgIUmSqjFYSJKkagwWkiSpGoOFJEmq\nxmAhSZKqMVhIkqRqDBaSJKma/w/+di8H7bluGQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f8f0345e810>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.hist(durations,bins=50)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.1917304108333309"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "durations = np.array(durations)\n",
    "durations[durations>10].sum()/3600"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "5601it [00:17, 318.71it/s]\n"
     ]
    }
   ],
   "source": [
    "# Remake data using Kyubyong method\n",
    "ref_len = 10. * sr # We controls the length of each clip such that it is close to this.\n",
    "min_len = 1. * sr # We ignore a split point if it causes an audio clip under 5 1 sec.\n",
    "\n",
    "# Make book directory\n",
    "os.system('mkdir '+LANGUAGE_CODE+'/'+BOOK_TITLE+'_Kyubyong')\n",
    "\n",
    "# Save durations\n",
    "durations_k = []\n",
    "\n",
    "# Create new splitpoints\n",
    "cursor, _nsamples, filtered = 0, 0, []\n",
    "for s in [int(sil*sr) for sil in silences] :\n",
    "    nsamples = s - cursor\n",
    "    if nsamples > ref_len and _nsamples > min_len:\n",
    "        cursor = _s\n",
    "        filtered.append(cursor/sr)\n",
    "        nsamples = s - cursor\n",
    "\n",
    "    _nsamples, _s = nsamples, s\n",
    "\n",
    "# Split data and write\n",
    "for i,s in tqdm(enumerate(filtered)):\n",
    "    if i == 0:\n",
    "        prev_cursor = 0\n",
    "    curr_cursor = int(sr*s)\n",
    "    clip = final[prev_cursor:curr_cursor]\n",
    "    output_num = str(i).zfill(4)\n",
    "    librosa.output.write_wav(LANGUAGE_CODE+'/'+BOOK_TITLE+'_Kyubyong'+'/'+BOOK_TITLE+output_num,clip,sr)\n",
    "    prev_cursor = curr_cursor\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "5601it [00:00, 595842.87it/s]\n"
     ]
    }
   ],
   "source": [
    "# Get statistics\n",
    "durations_k = []\n",
    "for i,s in tqdm(enumerate(filtered)):\n",
    "    if i == 0:\n",
    "        prev_cursor = 0\n",
    "    curr_cursor = s\n",
    "    durations_k.append(curr_cursor-prev_cursor)\n",
    "    prev_cursor = curr_cursor"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAhYAAAFkCAYAAAB8RXKEAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAAPYQAAD2EBqD+naQAAH3lJREFUeJzt3X+U3XV95/HnCzBJQzehNpKU2pzCoc1OT/eoGQpyLGw1\nnrLKluqyp2U05RS266pA2dlti57iKSV7uopHQhHscipbRXR6MKwHaymI1KUWkZwStFsdskuFXhAS\ne0tIUsLwI/nsH/cbvbnOTDKTz525kzwf59yTmc/nPd/7uZ98Z+Y1n/v53ptSCpIkSTUcM98DkCRJ\nRw6DhSRJqsZgIUmSqjFYSJKkagwWkiSpGoOFJEmqxmAhSZKqMVhIkqRqDBaSJKkag4UkSapmxsEi\nyVlJPp/kO0n2JTlvmtr/0dT8Zk/74iQ3Jmkn2Z1kU5ITe2p+JMmnk+xMsiPJx5McP9PxSpKkuTOb\nFYvjga8D7wWmfKORJG8HzgC+M0n3dcC5wPnA2cBJwO09NZ8BhoB1Te3ZwE2zGK8kSZojOZw3IUuy\nD3hbKeXzPe0/DjwAnAPcCWwspVzf9C0D/hG4oJTyuaZtDTAOvL6UsjnJEPBNYLiU8nBTcw7w58Cr\nSynbZj1oSZLUN9X3WCQJcAtwTSllfJKSYeA44N79DaWUrUALOLNpej2wY3+oaHyJzgrJGbXHLEmS\n6jiuD8d8H/BiKeWGKfpXNf27etq3N337a77b3VlK2Zvkma6aAyT5UTorJI8DE7MbuiRJR6UlwE8C\nd5dS/ulwDlQ1WCQZBn4TeF3N4x6ic4BPz8P9SpJ0pHgnnT2Os1Z7xeLngVcBT3SeEQHgWODaJP+5\nlHIKsA1YlGRZz6rFyqaP5t/eq0SOBV7ZVdPrcYBbb72VoaGhCg/l6DE6OsrGjRvnexh9Nz4+zvr1\n64ENwMlTVD0GfOCg59HRMme1OW8z55zNjvM2M9//+dj5XXo4ageLW4B7etq+2LT/SfP5Q8DLdK72\n6N68uZrOhk+af09I8rqufRbrgAAPTnHfEwBDQ0OsXbv28B/JUWT58uVH2Zy9FZjq8W4BPnDQ8+jo\nm7M6nLeZc85mx3mbtcPeSjDjYNG8lsSpdH7JA5yS5DXAM6WUJ4AdPfUvAdtKKf8PoJSyK8nNdFYx\ndgC7geuB+0spm5uaR5LcDfxxkvcAi4CPAmNeESJJ0uCazYrFacCX6VyhUYCPNO2fBC6epH6y61lH\ngb3AJmAxcBdwSU/NO4Ab6FwNsq+pvXwW45UkSXNkxsGilHIfM7hMtdlX0dv2AnBZc5vq654F1s90\nfJIkaf74XiFiZGRkvoew4Dhns+O8zZxzNjvO2/wxWMhvwFlwzmbHeZs552x2nLf5Y7CQJEnVGCwk\nSVI1BgtJklSNwUKSJFVjsJAkSdUYLCRJUjUGC0mSVI3BQpIkVWOwkCRJ1RgsJElSNQYLSZJUjcFC\nkiRVY7CQJEnVGCwkSVI1BgtJklSNwUKSJFVjsJAkSdUYLCRJUjUGC0mSVI3BQpIkVWOwkCRJ1Rgs\nJElSNQYLSZJUjcFCkiRVY7CQJEnVGCwkSVI1BgtJklSNwUKSJFVjsJAkSdUYLCRJUjUGC0mSVM2M\ng0WSs5J8Psl3kuxLcl5X33FJPpTkb5P8c1PzySQ/1nOMxUluTNJOsjvJpiQn9tT8SJJPJ9mZZEeS\njyc5fvYPVZIk9dtsViyOB74OvBcoPX1LgdcCvw+8Dng7sAa4o6fuOuBc4HzgbOAk4Paems8AQ8C6\npvZs4KZZjFeSJM2R42b6BaWUu4C7AJKkp28XcE53W5JLgQeTvLqU8mSSZcDFwAWllPuamouA8SSn\nl1I2JxlqjjNcSnm4qbkM+PMkv1VK2TbjRypJkvpuLvZYnEBnZePZ5vNhOoHm3v0FpZStQAs4s2l6\nPbBjf6hofKk5zhn9HrAkSZqdvgaLJIuBDwKfKaX8c9O8CnixWd3otr3p21/z3e7OUspe4JmuGkmS\nNGD6FiySHAd8ls4qw3v7dT+SJGlwzHiPxaHoChU/Abypa7UCYBuwKMmynlWLlU3f/preq0SOBV7Z\nVTOp0dFRli9ffkDbyMgIIyMjs3kokiQdUcbGxhgbGzugbefOndWOXz1YdIWKU4A3llJ29JQ8BLxM\n52qPzzVfswZYDTzQ1DwAnJDkdV37LNYBAR6c7v43btzI2rVrazwUSZKOOJP9sb1lyxaGh4erHH/G\nwaJ5LYlT6fySBzglyWvo7H94ms5lo68F/i3wiiQrm7pnSikvlVJ2JbkZuDbJDmA3cD1wfyllM0Ap\n5ZEkdwN/nOQ9wCLgo8CYV4RIkjS4ZrNicRrwZTp7Jwrwkab9k3Rev+KXmvavN+1pPn8j8FdN2yiw\nF9gELKZz+eolPffzDuAGOleD7GtqL5/FeCVJ0hyZzetY3Mf0mz4PuiG0lPICcFlzm6rmWWD9TMcn\nSZLmj+8VIkmSqjFYSJKkagwWkiSpGoOFJEmqxmAhSZKqMVhIkqRqDBaSJKmavrxXiNQPrVaLdrs9\nZf+KFStYvXr1HI5IktTLYKEFodVqsWbNEBMTe6asWbJkKVu3jhsuJGkeGSy0ILTb7SZU3AoMTVIx\nzsTEetrttsFCkuaRwUILzBDgu9dK0qBy86YkSarGYCFJkqoxWEiSpGoMFpIkqRqDhSRJqsZgIUmS\nqjFYSJKkagwWkiSpGoOFJEmqxmAhSZKqMVhIkqRqDBaSJKkag4UkSarGYCFJkqoxWEiSpGoMFpIk\nqRqDhSRJqsZgIUmSqjFYSJKkagwWkiSpGoOFJEmqxmAhSZKqMVhIkqRqZhwskpyV5PNJvpNkX5Lz\nJqm5OslTSfYkuSfJqT39i5PcmKSdZHeSTUlO7Kn5kSSfTrIzyY4kH09y/MwfoiRJmiuzWbE4Hvg6\n8F6g9HYmuQK4FHgXcDrwHHB3kkVdZdcB5wLnA2cDJwG39xzqM8AQsK6pPRu4aRbjlSRJc+S4mX5B\nKeUu4C6AJJmk5HJgQynlC03NhcB24G3AbUmWARcDF5RS7mtqLgLGk5xeStmcZAg4BxgupTzc1FwG\n/HmS3yqlbJvpuCVJUv9V3WOR5GRgFXDv/rZSyi7gQeDMpuk0OoGmu2Yr0OqqeT2wY3+oaHyJzgrJ\nGTXHLEmS6qm9eXMVnV/+23vatzd9ACuBF5vAMVXNKuC73Z2llL3AM101kiRpwMz4qZBBNzo6yvLl\nyw9oGxkZYWRkZJ5GJEnS4BgbG2NsbOyAtp07d1Y7fu1gsQ0InVWJ7lWLlcDDXTWLkizrWbVY2fTt\nr+m9SuRY4JVdNZPauHEja9eunfUDkCTpSDbZH9tbtmxheHi4yvGrPhVSSnmMzi/+dfvbms2aZwBf\nbZoeAl7uqVkDrAYeaJoeAE5I8rquw6+jE1oerDlmSZJUz4xXLJrXkjiVzi95gFOSvAZ4ppTyBJ1L\nSa9M8ijwOLABeBK4AzqbOZPcDFybZAewG7geuL+UsrmpeSTJ3cAfJ3kPsAj4KDDmFSGSJA2u2TwV\nchrwZTqbNAvwkab9k8DFpZRrkiyl85oTJwBfAd5SSnmx6xijwF5gE7CYzuWrl/TczzuAG+hcDbKv\nqb18FuOVJElzZDavY3EfB3kKpZRyFXDVNP0vAJc1t6lqngXWz3R8kiRp/vheIZIkqRqDhSRJqsZg\nIUmSqjFYSJKkagwWkiSpGoOFJEmqxmAhSZKqMVhIkqRqDBaSJKkag4UkSarGYCFJkqoxWEiSpGoM\nFpIkqRqDhSRJqsZgIUmSqjFYSJKkagwWkiSpGoOFJEmqxmAhSZKqMVhIkqRqDBaSJKkag4UkSarG\nYCFJkqoxWEiSpGoMFpIkqRqDhSRJqsZgIUmSqjFYSJKkagwWkiSpGoOFJEmqxmAhSZKqOW6+B6Aj\nX6vVot1uT1uzYsUKVq9ePUcjkiT1i8FCfdVqtVizZoiJiT3T1i1ZspStW8cNF5K0wFV/KiTJMUk2\nJPl2kj1JHk1y5SR1Vyd5qqm5J8mpPf2Lk9yYpJ1kd5JNSU6sPV71V7vdbkLFrcBDU9xuZWJiz0FX\nNSRJg68fKxbvA/4TcCHwLeA04BNJni2l3ACQ5Arg0qbmceC/AXcnGSqlvNgc5zrgLcD5wC7gRuB2\n4Kw+jFl9NwSsne9BSJL6rB/B4kzgjlLKXc3nrSTvAE7vqrkc2FBK+QJAkguB7cDbgNuSLAMuBi4o\npdzX1FwEjCc5vZSyuQ/jliRJh6kfV4V8FViX5KcAkrwGeANwZ/P5ycAq4N79X1BK2QU8SCeUQGeV\n47iemq1Aq6tGkiQNmH6sWHwQWAY8kmQvnfDyu6WUP236VwGFzgpFt+1NH8BK4MUmcExVI0mSBkw/\ngsWvAu8ALqCzx+K1wB8meaqU8qk+3N8BRkdHWb58+QFtIyMjjIyM9PuuJUkaeGNjY4yNjR3QtnPn\nzmrH70ewuAb476WUzzaffzPJTwLvBz4FbANCZ1Wie9ViJfBw8/E2YFGSZT2rFiubvilt3LiRtWvd\nJChJ0mQm+2N7y5YtDA8PVzl+P/ZYLAX29rTt239fpZTH6ISDdfs7m82aZ9DZnwGdaxBf7qlZA6wG\nHujDmCVJUgX9WLH4M+DKJE8C36RzjeEo8PGumuuamkfpXG66AXgSuAM6mzmT3Axcm2QHsBu4Hrjf\nK0IkSRpc/QgWl9IJCjcCJwJPAX/UtAFQSrkmyVLgJuAE4CvAW7pewwI6YWQvsAlYDNwFXNKH8UqS\npEqqB4tSynPAf2lu09VdBVw1Tf8LwGXNTZIkLQC+u6kkSarGYCFJkqoxWEiSpGoMFpIkqRqDhSRJ\nqqYfl5tKR7RWq0W73Z62ZsWKFaxevXqORiRJg8NgIc1Aq9VizZohJib2TFu3ZMlStm4dN1xIOuoY\nLKQZaLfbTai4FRiaomqciYn1tNttg4Wko47BQpqVITqvVi9J6ubmTUmSVI3BQpIkVWOwkCRJ1Rgs\nJElSNQYLSZJUjcFCkiRVY7CQJEnVGCwkSVI1BgtJklSNwUKSJFVjsJAkSdUYLCRJUjUGC0mSVI3B\nQpIkVWOwkCRJ1RgsJElSNQYLSZJUjcFCkiRVY7CQJEnVGCwkSVI1BgtJklSNwUKSJFVjsJAkSdUY\nLCRJUjV9CRZJTkryqSTtJHuSfCPJ2p6aq5M81fTfk+TUnv7FSW5sjrE7yaYkJ/ZjvJIkqY7qwSLJ\nCcD9wAvAOcAQ8F+BHV01VwCXAu8CTgeeA+5OsqjrUNcB5wLnA2cDJwG31x6vJEmq57g+HPN9QKuU\n8htdbf/QU3M5sKGU8gWAJBcC24G3AbclWQZcDFxQSrmvqbkIGE9yeillcx/GLUmSDlM/ngr5JeBv\nktyWZHuSLUm+FzKSnAysAu7d31ZK2QU8CJzZNJ1GJ/R012wFWl01kiRpwPQjWJwCvAfYCvwi8EfA\n9Ul+relfBRQ6KxTdtjd9ACuBF5vAMVWNJEkaMP14KuQYYHMp5QPN599I8rPAu4FP9eH+JEnSgOhH\nsHgaGO9pGwf+XfPxNiB0ViW6Vy1WAg931SxKsqxn1WJl0zel0dFRli9ffkDbyMgIIyMjM3kMkiQd\nkcbGxhgbGzugbefOndWO349gcT+wpqdtDc0GzlLKY0m2AeuAvwVoNmueAdzY1D8EvNzUfK6pWQOs\nBh6Y7s43btzI2rVrpyuRJOmoNdkf21u2bGF4eLjK8fsRLDYC9yd5P3AbncDwG8B/7Kq5DrgyyaPA\n48AG4EngDuhs5kxyM3Btkh3AbuB64H6vCJEkaXBVDxallL9J8nbgg8AHgMeAy0spf9pVc02SpcBN\nwAnAV4C3lFJe7DrUKLAX2AQsBu4CLqk9XkmSVE8/ViwopdwJ3HmQmquAq6bpfwG4rLlJkqQFwPcK\nkSRJ1RgsJElSNQYLSZJUjcFCkiRVY7CQJEnVGCwkSVI1BgtJklSNwUKSJFVjsJAkSdUYLCRJUjUG\nC0mSVE1f3itER45Wq0W73Z62ZsWKFaxevXqORiRJGmQGC02p1WqxZs0QExN7pq1bsmQpW7eOGy4k\nSQYLTa3dbjeh4lZgaIqqcSYm1tNutw0WkiSDhQ7FELB2vgchSVoA3LwpSZKqMVhIkqRqDBaSJKka\ng4UkSarGYCFJkqoxWEiSpGoMFpIkqRqDhSRJqsZgIUmSqjFYSJKkagwWkiSpGoOFJEmqxmAhSZKq\nMVhIkqRqDBaSJKkag4UkSarGYCFJkqoxWEiSpGr6HiySvC/JviTX9rRfneSpJHuS3JPk1J7+xUlu\nTNJOsjvJpiQn9nu8kiRp9voaLJL8HPAu4Bs97VcAlzZ9pwPPAXcnWdRVdh1wLnA+cDZwEnB7P8cr\nSZIOT9+CRZIfBm4FfgN4tqf7cmBDKeULpZS/Ay6kExze1nztMuBiYLSUcl8p5WHgIuANSU7v15gl\nSdLh6eeKxY3An5VS/rK7McnJwCrg3v1tpZRdwIPAmU3TacBxPTVbgVZXjSRJGjDH9eOgSS4AXksn\nIPRaBRRge0/79qYPYCXwYhM4pqqRJEkDpnqwSPJqOvsj3lxKean28SVJ0uDqx4rFMPAqYEuSNG3H\nAmcnuRT4l0DorEp0r1qsBB5uPt4GLEqyrGfVYmXTN6XR0VGWL19+QNvIyAgjIyOzfDiSJB05xsbG\nGBsbO6Bt586d1Y7fj2DxJeBf9bR9AhgHPlhK+XaSbcA64G/he5s1z6CzLwPgIeDlpuZzTc0aYDXw\nwHR3vnHjRtauXVvlgUiSdKSZ7I/tLVu2MDw8XOX41YNFKeU54FvdbUmeA/6plDLeNF0HXJnkUeBx\nYAPwJHBHc4xdSW4Grk2yA9gNXA/cX0rZXHvMkiSpjr5s3pxEOeCTUq5JshS4CTgB+ArwllLKi11l\no8BeYBOwGLgLuGRuhitJkmZjToJFKeVNk7RdBVw1zde8AFzW3CRJ0gLge4VIkqRqDBaSJKkag4Uk\nSarGYCFJkqoxWEiSpGoMFpIkqRqDhSRJqsZgIUmSqpmrV97UAGq1WrTb7Sn7x8fHp+yTJGkyBouj\nVKvVYs2aISYm9sz3UCRJRxCDxVGq3W43oeJWYGiKqjuBD8zdoCRJC57B4qg3BEz1NvM+FSJJmhk3\nb0qSpGoMFpIkqRqDhSRJqsZgIUmSqjFYSJKkagwWkiSpGoOFJEmqxmAhSZKqMVhIkqRqDBaSJKka\ng4UkSarGYCFJkqoxWEiSpGoMFpIkqRqDhSRJqsZgIUmSqjFYSJKkagwWkiSpGoOFJEmqxmAhSZKq\nMVhIkqRqDBaSJKma6sEiyfuTbE6yK8n2JJ9L8tOT1F2d5Kkke5Lck+TUnv7FSW5M0k6yO8mmJCfW\nHq8kSaqnHysWZwEfBc4A3gy8Avhikh/aX5DkCuBS4F3A6cBzwN1JFnUd5zrgXOB84GzgJOD2PoxX\nkiRVclztA5ZS3tr9eZJfB74LDAN/3TRfDmwopXyhqbkQ2A68DbgtyTLgYuCCUsp9Tc1FwHiS00sp\nm2uPW5IkHb7qwWISJwAFeAYgycnAKuDe/QWllF1JHgTOBG4DTmvG1l2zNUmrqTnqg0Wr1aLdbk9b\ns2LFClavXj1HI5Ikqc/BIknoPKXx16WUbzXNq+gEje095dubPoCVwIullF3T1By1Wq0Wa9YMMTGx\nZ9q6JUuWsnXruOFCkjRn+r1i8THgZ4A39Pl+vmd0dJTly5cf0DYyMsLIyMhcDaHv2u12EypuBYam\nqBpnYmI97XbbYCFJ+p6xsTHGxsYOaNu5c2e14/ctWCS5AXgrcFYp5emurm1A6KxKdK9arAQe7qpZ\nlGRZz6rFyqZvShs3bmTt2rWHO/wFYgg4Wh6rJKmGyf7Y3rJlC8PDw1WO35fXsWhCxS8DbyyltLr7\nSimP0QkH67rql9G5iuSrTdNDwMs9NWuA1cAD/RizJEk6fNVXLJJ8DBgBzgOeS7Ky6dpZSploPr4O\nuDLJo8DjwAbgSeAO+N5mzpuBa5PsAHYD1wP3e0WIJEmDqx9PhbybzubM/93TfhFwC0Ap5ZokS4Gb\n6Fw18hXgLaWUF7vqR4G9wCZgMXAXcEkfxitJkirpx+tYHNLTK6WUq4Crpul/AbisuUmSpAXA9wqR\nJEnVGCwkSVI1BgtJklSNwUKSJFVjsJAkSdUYLCRJUjUGC0mSVI3BQpIkVWOwkCRJ1RgsJElSNQYL\nSZJUjcFCkiRVY7CQJEnVGCwkSVI1BgtJklSNwUKSJFVjsJAkSdUcN98DOBq1Wi3a7faU/StWrGD1\n6tVzOCL1w/j4+LT9/j9LOhIZLOZYq9VizZohJib2TFmzZMlStm4d95fOgvU0cAzr16+ftsr/Z0lH\nIoPFHGu3202ouBUYmqRinImJ9bTbbX/hLFjPAvuY+v8Y/H+WdKQyWMybIWDtfA9CfeX/saSjj5s3\nJUlSNQYLSZJUjcFCkiRV4x4LaUAd7LJk8JJVSYPHYCENoEO5LBm8ZFXS4DFYSAPo4Jclg5esShpE\nBgtpoHnJqqSFxWAxAz7nLUnS9AwWh8jnvLVQ1QjEhmpJh8pgcYh8zlsLUY1AbKiWNBMGixnzOW8t\nHDUCsaFa0kz4AlkCxuZ7AAvQXfM9gBnaH4gnu00VFuofY2zMc22mnLPZcd7mz8AHiySXJHksyfNJ\nvpbk5+Z7TEcevwFn7u75HsCC5A/7mXPOZsd5mz8D/VRIkl8FPgK8C9gMjAJ3J/npUsr0O8kkzbnx\n8fFp+59//vk5Gomk+TLQwYJOkLiplHILQJJ3A+cCFwPXzOfAJHV7GjiG9evXT1t1zDHH0mq1ptyH\nUevqk4MdxytYpP4Z2GCR5BXAMPAH+9tKKSXJl4Aza97XCy+8wK/92kU88sjWKWuef376HfHS0e1Z\nYB8H2+C5b9/UGzxrXX1yKMeZqytYvExXR6OBDRbACuBYYHtP+3ZgzST1S+DgS7GTeeKJJ/jsZ8eA\nlcAPTVG1/4fDncBU9/FYp+LOO6ccx2OPPXaQ49Q4xsGPc+AxngQ+Pckx7q98P0fSvG1n8jmDhTVv\ntcay/zE/NkX/od1PJwz8B+DHpjjG00xM3Mwtt9zCySefPPm9HPQ4Bz8GwDHHHMO+ffumeTzT17Tb\nbX77t9/HSy9NTHuMV7xiMR/+8IdYsWLFpP1PPvkkY2Nj047lcMd6qDUL6X4Gad4OpabW/axYsYJX\nvepV09ZMput7csmMv7hHSimHe4y+SPJjwHeAM0spD3a1fwg4u5RyZk/9O5j6J70kSTq4d5ZSPnM4\nBxjkFYs2sJfOMkK3lcC2ServBt4JPA5M/yeCJEnqtgT4SSpc8jawKxYASb4GPFhKubz5PEALuL6U\n8uF5HZwkSfoBg7xiAXAt8IkkD/H9y02XAp+Yz0FJkqTJDXSwKKXclmQFcDWdp0C+DpxTSvnH+R2Z\nJEmazEA/FSJJkhaWgX9Jb0mStHAYLCRJUjVHRLDwjcpmJsnvJdnXc/vWfI9r0CQ5K8nnk3ynmaPz\nJqm5OslTSfYkuSfJqfMx1kFxsDlL8ieTnHt3ztd4B0GS9yfZnGRXku1JPpfkpyep81zrcijz5vl2\noCTvTvKNJDub21eT/JuemsM+zxZ8sOh6o7LfA14HfIPOG5VN/jJ22u/v6GyIXdXcfn5+hzOQjqez\nYfi9wA9sRkpyBXApnTfJOx14js65t2guBzlgpp2zxl9w4Lk3MjdDG1hnAR8FzgDeDLwC+GKS770M\nsOfapA46bw3Pt+97ArgCWEvnLTP+ErgjyRBUPM9KKQv6BnwN+MOuz0PnNap/Z77HNqg3OiFsy3yP\nYyHd6LwRxnk9bU8Bo12fLwOeB35lvsc7CLcp5uxPgP8132Mb5BudtzPYB/x8V5vn2uzmzfPt4PP2\nT8BFzcdVzrMFvWLR9UZl9+5vK53ZqP5GZUegn2qWq/8+ya1JfmK+B7SQJDmZzl8/3efeLuBBPPcO\n5heapetHknwsySvne0AD5gQ6qz3PgOfaDBwwb1083yaR5JgkF9B5baiv1jzPFnSwYPo3Kls198NZ\nML4G/DpwDvBu4GTgr5IcP5+DWmBW0fkh5rk3M38BXAi8Cfgd4F8DdzavqnvUa+bhOuCvSyn79z15\nrh3EFPMGnm8/IMnPJtkNvAB8DHh7KWUrFc+zgX6BLPVHKaX7teD/Lslm4B+AX6GzdCj1RSnltq5P\nv5nk/wB/D/wC8OV5GdRg+RjwM8Ab5nsgC8yk8+b5NqlHgNcAy4F/D9yS5Oyad7DQVyxm+kZlmkQp\nZSfwf4Gjepf5DG2js5/Hc+8wlFIeo/N9fNSfe0luAN4K/EIp5emuLs+1aUwzbz/A8w1KKS+XUr5d\nSnm4lPK7dC54uJyK59mCDhallJeAh4B1+9uaJa51wFfna1wLTZIfpvONNu03pb6v+QG1jQPPvWV0\ndqh77h2iJK8GfpSj/Nxrfjn+MvDGUkqru89zbWrTzdsU9Z5vP+gYYHHN8+xIeCrENyqboSQfBv6M\nztMfPw78PvASMDaf4xo0zZ6TU+mkeIBTkrwGeKaU8gSd53SvTPIo8Diwgc4VSXfMw3AHwnRz1tx+\nD7idzg+wU4EP0VktO+y3al6oknyMziWQ5wHPJdn/F+POUspE87HnWo+DzVtzLnq+dUnyB3T2nbSA\nfwG8k86+k19sSuqcZ/N9qUuly2Xe20zC88ADwGnzPaZBvtEJEE8289UCPgOcPN/jGrRb8w23j87T\nbd23/9lVcxWdS7T20Plhdep8j3tQ5wxYAtxF54f8BPBt4I+AV833uOd5ziabr73AhT11nmszmDfP\nt0nn7OPNPDzfzMsXgTf11Bz2eeabkEmSpGoW9B4LSZI0WAwWkiSpGoOFJEmqxmAhSZKqMVhIkqRq\nDBaSJKkag4UkSarGYCFJkqoxWEiSpGoMFpIkqRqDhSRJqub/A5yTpnitsZMyAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f8f02010310>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.hist(durations_k,bins=50)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "4619it [00:09, 488.59it/s]\n"
     ]
    }
   ],
   "source": [
    "# tommy style: just center around 5 sec\n",
    "\n",
    "ref_len = 12. # We controls the length of each clip such that it is close to this.\n",
    "min_len = 1. # We ignore a split point if it causes an audio clip under 5 1 sec.\n",
    "\n",
    "# Make book directory\n",
    "os.system('mkdir '+LANGUAGE_CODE+'/'+BOOK_TITLE+'_Tommy')\n",
    "\n",
    "\n",
    "# Create new splitpoints\n",
    "filtered = []\n",
    "start_cursor = 0\n",
    "prev_cursor = 0\n",
    "for s in silences:\n",
    "    cur_len = s-start_cursor\n",
    "    if cur_len > ref_len:\n",
    "        filtered.append(prev_cursor)\n",
    "        start_cursor = prev_cursor\n",
    "    prev_cursor = s\n",
    "\n",
    "\n",
    "\n",
    "# cursor, _nsamples, filtered, _s = 0, 0, [], 0\n",
    "# for s in [int(sil*sr) for sil in silences] :\n",
    "#     nsamples = s - cursor\n",
    "#     if nsamples > ref_len:\n",
    "#         cursor = _s\n",
    "#         filtered.append(cursor/sr)\n",
    "#         nsamples = s - cursor\n",
    "\n",
    "#     _nsamples, _s = nsamples, s\n",
    "\n",
    "# Split data and write\n",
    "for i,s in tqdm(enumerate(filtered)):\n",
    "    if i == 0:\n",
    "        prev_cursor = 0\n",
    "    curr_cursor = int(sr*s)\n",
    "    clip = final[prev_cursor:curr_cursor]\n",
    "    output_num = str(i).zfill(4)\n",
    "    librosa.output.write_wav(LANGUAGE_CODE+'/'+BOOK_TITLE+'_Tommy'+'/'+BOOK_TITLE+output_num,clip,sr)\n",
    "    prev_cursor = curr_cursor"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "4619it [00:00, 579610.78it/s]\n"
     ]
    }
   ],
   "source": [
    "# Get statistics\n",
    "durations_t = []\n",
    "for i,s in tqdm(enumerate(filtered)):\n",
    "    if i == 0:\n",
    "        prev_cursor = 0\n",
    "    curr_cursor = s\n",
    "    durations_t.append(curr_cursor-prev_cursor)\n",
    "    prev_cursor = curr_cursor"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAg4AAAFkCAYAAABIPLOYAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n5VwAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f8f016ffc90>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.hist(durations_t,bins=50)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "durations_t = np.array(durations_t)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "7.6407373862856796"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "durations_t[durations_t>10].sum()/3500"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [conda root]",
   "language": "python",
   "name": "conda-root-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}