Voice Activity Detector Prototype.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import scipy.signal as signal\n",
    "import sounddevice as sd\n",
    "import soundfile as sf\n",
    "import scipy.io.wavfile as wavfile\n",
    "import matplotlib.pyplot as plt\n",
    "import pickle\n",
    "from sklearn.svm import SVC\n",
    "import sklearn.metrics as metrics\n",
    "from sklearn.model_selection import StratifiedShuffleSplit\n",
    "from sklearn.model_selection import GridSearchCV"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Define Basic Voice Activation Detector"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "class VoiceActivityDetector():   \n",
    "    def __init__(self, fs):\n",
    "        self.sample_rate = fs\n",
    "        self.sample_rate = 44100\n",
    "        self.speech_on = False\n",
    "    \n",
    "        #Spectrum Based:\n",
    "        self.speech_energy_threshold = 0.6 #60% of energy in voice band\n",
    "        self.speech_start_band = 300\n",
    "        self.speech_end_band = 3000\n",
    "    \n",
    "    def CheckActivation(self, data):\n",
    "        speech_band_energy_fraction = self.GetSpeechBandEnergyFraction(data)\n",
    "        self.speech_on = self.ClassifySpeech(speech_band_energy_fraction)\n",
    "    \n",
    "    def FreqInRange(self, f):\n",
    "        return freq[i] > self.speech_start_band and freq[i] < self.speech_end_band\n",
    "    \n",
    "    def GetSpeechBandEnergyFraction(self,data):\n",
    "        freq, power = signal.periodogram(data, self.sample_rate)\n",
    "        total_energy = np.sum(power)\n",
    "        in_range = lambda f : (f > self.speech_start_band) and (f < self.speech_end_band)\n",
    "        band_energy = sum([power[i] for i in range(len(power)) if in_range(freq[i])]) \n",
    "        return band_energy / total_energy\n",
    "    \n",
    "    def ClassifySpeech(self, speech_band_energy_fraction):\n",
    "        return (speech_band_energy_fraction > self.speech_energy_threshold)\n",
    "\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "[fs, data] = wavfile.read('wav-sample.wav')\n",
    "VAD = VoiceActivityDetector(fs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "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(data)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.5657182945982169"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "VAD.GetSpeechBandEnergyFraction(data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXwAAAD7CAYAAABpJS8eAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjMsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+AADFEAAAgAElEQVR4nO2deZhcVZn/v6f2tfct3Z1OZycJgQSasAVkN7KqwyiIIoowjD8cHHRmYNBRRMRlRkdHRkEFxhkFGQeVQDCETfaQEALZk07SSbrT6b26umtfzu+Pe8+te6uru6u6tltd7+d5+knVrVu3zu1Uf+97v+c978s45yAIgiBmP4ZiD4AgCIIoDCT4BEEQZQIJPkEQRJlAgk8QBFEmkOATBEGUCST4BEEQZQIJPkEQRJlAgk8QBFEmmAr1QYyxBQDuAVDJOb82nffU1dXx9vb2vI6LIAhitvHuu+8Ocs7rk7dnJfiMsUcAXAmgn3N+smr7OgA/BmAE8EvO+Xc554cA3MwY+326x29vb8fWrVuzGSJBEETZwRg7kmp7tpbOYwDWJX2QEcCDAD4CYDmA6xljy7P8HIIgCCJLshJ8zvmrAIaTNq8B0Mk5P8Q5DwN4AsA16R6TMXYrY2wrY2zrwMBANsMjCIIgVORj0rYFwDHV824ALYyxWsbYzwGsZozdPdmbOecPc847OOcd9fUTLCiCIAhihhRs0pZzPgTgtnT2ZYxdBeCqRYsW5XdQBEEQZUQ+IvweAHNVz1vlbWnDOV/POb+1srIypwMjCIIoZ/Ih+FsALGaMzWeMWQBcB+DpTA7AGLuKMfbw6OhoHoZHEARRnmQl+IyxxwG8BWApY6ybMXYz5zwK4HYAGwHsAfAk53xXJselCJ8gCCL3ZOXhc86vn2T7BgAbsjl2KXBs2I/OgXFcuLSh2EMhCIKYFl2WVigVS+fSH/0Fn3t0S7GHQRAEkRa6FPxSsXSCkTgAgPoCEwRRCuhS8EuNSIwEnyAI/aNLwS8VS0cQiMSKPQSCIIhp0aXgl4qlIwiR4BMEUQLoUvBLDeHlEwRB6BldCj5ZOgRBELlHl4JfapZOkASfIIgSQJeCX2qQ4BMEUQqQ4M8Qde49WToEQZQCJPgzRD1RS5O2BEGUAroU/FKYtPWHo8pjsnQIgigFdCn4pTBp6w8nRJ4EnyCIUkCXgl8KqH17EnyCIEoBEvwZElL59j2eQBFHQhAEkR4k+DMkHEtE9b947TD6vMEijoYgCGJ6dCn4pTBpG4pqM3P2nRgr0kgIgiDSQ5eCXwqTtmFZ8L/zsZUAgMODvmIOhyAIYlp0Kfh6ZMOOXuw94VWeiwj/lNZKuKwmEnyCIHRPVj1ty4kv/mYbAKDru1cASET4VpMB82odODJEgk8QhL6hCH+GJATfCJfVpMnLJwiC0CMk+GkQj09sYSgsHYvJAIvJgHCMyisQBKFvSPDTIBidGL2H5W0WkwFWkxGhSBzhaBzvHR1JeYEgCIIoNiT4aRBQ2TVD4yHE41yJ6K0mA6xyhP/sjuP42H++iW+u31WsoRIEQUyKLgVfb3n46jIKp3/7Bfzftm5lpa1FFvxQNIah8TAAYGePPsZNEAShRpeCr7c8/ORaOfv7xhCOxcEYYDIwWM0GhCJxZT+jgaV1XI8/rKmrTxAEkU90Kfh6IxDWTsjaLSaEo3FYTQYwxmAxSpaOqIufjob3e4NY9a1N+NlfDuZjyARBEBMgwU8DYek8etMZsJkN8AYiCEXjsBilX5/VLE3aiv3SydjpHwsBAH6/tTtPoyYIgtBCgp8GQsgr7CbUuawYFYJvMgKAEuErgh+dXvC9gQgAYDwUnWZPgiCI3ECCn4J7/rADV/zkNeW5yNKxmY2otJsxGogolg4gZerE4hw+WbzTifBH/JLg+0jwCYIoEFRaIQW/2XwUgLTgymBgymSsXSX4DotREXyL/O+oHLVHphD853edwA827sOVpzQDAHy0QpcgiAJBEf4UHB+VGpsIq8ZukQTf4w8jHI0rQm9NEvypLJ2Nu/pwoH8cP3phv7JtNBAha4cgiLxDgj8FBwekgmjC0klE+FGEY3FVhC95+YkIX5um899vH8Ha772ESCwOm3nir/zUe5/Hmfe/kLfzIAiCAAoo+IwxJ2Psvxhjv2CM3VCoz80UdV784YFxAIkI32Y2osZpgccfxngwOjHC90+M8Dnn+Pofd6J7JIDRQATeYBQtVfYJn0vWDkEQ+SYrwWeMPcIY62eM7Uzavo4xto8x1skYu0ve/HEAv+ec3wLg6mw+N5+orRUxsRqKxMCYJOwrWyoRjXNsPTKCpkpJuK3mJEtH5eF7g4nj+UJRjAUjqHVZcP6S+ryfS644NuzHp3+5WTk/giBKk2wj/McArFNvYIwZATwI4CMAlgO4njG2HEArgGPybroNZ0d8CVETAhdSLbI6bV618vqyOW4AUPLxo3LRtHA0rtwpqC8g46EovIEIKmxm/Prza7D+9rWaz04nnbMYvN45iNc7B9HZT20cCaKUyUrwOeevAhhO2rwGQCfn/BDnPAzgCQDXAOiGJPpZf24+GfKFlMdeleALUW+ssCmvL5tTAUBaeJWM8PHHNRF+DN5gFG6blBxV5TBr3jPiD+fiFHJOl9zchWr+E0Rpkw/hbUEikgckoW8B8BSAv2KM/QzA+snezBi7lTG2lTG2dWBgIA/Dmxp1RJ6I8GMaUf/DF8/Bmvk1OF2O9sXFQI1IzVQfz6eK8AGg2mnRvEcUX9MbXXL7Rl+IBJ8gSpmC5eFzzn0APpfGfg8DeBgAOjo6Cl5ZTAi1y2pKCH4kkZEDAKvbqvHk35ytPLeqMm+kyplSbXynVbuwajwUxVgwigq79Gt3WrR3BsO+4gr+9mMeVNrNmF/n1Gw/MuQHAAQilDpKEKVMPiL8HgBzVc9b5W1pU8zyyOGodI2pc1kSgh/TCn4y6tcq7VL0Li4casH3+MMIRGJKhM+YVFXz1LlVALR2UjH46INv4MJ/fUWzjXOuWDoU4RNEaZMPwd8CYDFjbD5jzALgOgBPZ3KAYpZHFkJd57LCo4nwJ/r0Apc1caNUI9s0ogWi2tI54Q1K+9sS+3d99wo89OnTAehTUPvHQkoV0AB5+ARR0mSblvk4gLcALGWMdTPGbuacRwHcDmAjgD0AnuScZ9QCqpgRfjSeEHy1h2+ZIsJ32xKTr7UuSfBTefjDcgZQ8sXDLls7gYj+BPWw7N8DNGlLEKVOVh4+5/z6SbZvALAhi+OuB7C+o6PjlpkeY6ZEZEunscKKcDSO8VBUScucDLcqYq91WgEkcvGFpWM0MHjkLByzUdsgxS5PCAfCxfPI1QvOOOdgjCEai+NAXyIV01/E8REEkT26LJ7GGLsKwFWLFi0q+GcLoW6XJy57RgIIR+OosJsnfY9ZlaUjLJ2wYunEYDEaUGE3KWmXyXcLFpMBJgMraoSvXiw2HorCbTPj2p+/he3HPDAbGexmI0X4BFHi6DIfXg8evshU6R7xTxvhq6l1ai0dXygKp9UIp9UEj7xy15wijbPYgupXzR8MyM1Zth/zAABaqx1w28wk+ARR4uhS8AuFLxRF/1hQsy1Z8Hs8gWk9fDU1romTtk6rCXazEWPyIqyUgm8xTuidW0j8qs8eTFoPYDUZYLcYydIhiBJHl4JfqEnbv/75W1hz/4sApNr3kVhcWSHbVGmD1WRA90hgQh7+VNQmWToefxjVDgtsZiO8QRHhT2xyLglq8QRfPX/gD0c1nn4szuEs8vgIgsgeXQp+oSyd3b1eAEA0FscXfr0Vi+95ThFqi9GAercVg2Mh2dKZPC1TTZ1LmrQVEf6wP4JqpwU2s0GJ8FOtzLWbjUVNe1SLeSga1xR9u3hZI+yW4o6PIIjs0aXgF5qB8RBe2tsPAAhGYjAbGRiTJiqD0RjC0VjaEb4olyDsGSnCNyuZOABgTnEsh8VY1ElbteAHIzHFx//KpUvw1cuWwGExwU8rbQmipNFllk6h6R1N+PiHB32Kx26To+50Jm0ZAziXhBtILFIa9kmWjrrt4WQefnEj/ISYh6JxZW7j9PZqmIwG2MwGZQEWQRCliS4Fv1BpmWYjQyTG0acS/AP944ogi8yZdAT/la9egGPDgUROfSSGSCyOsWAUNU6L4t+Lz03GbjZhxBfIxWnNiGRLR5SJFusKbKbcTCoP+8Kwm43KYjOCIAqHLi2dQnn4oqZN72hQWTylifAticyaVCWQ1cyrdWLt4jrYVIIv8u4lDz/x/lQXj6Jn6agFPxLDsBi7XMLZmqMI/7T7NuHan7+Z9XEIgsgcXQp+oRCplt5gRGOzWOQI3GYyKOUV0vXwpUYpQDAcU/LuJ3j4KSwdRxHz8DnneOKdo8rzUDQOj1y5s8ohzUlYTUaEsrwgxeQGMbuOe7M6Trr8aXsPNYcnCBVlLfgioo7GuMbDFpOqdotRaYKSbh6+mOwNRBKCX2W3TCv49iJO2h4cGMe2ox5cfFIDAOn3MuKPwGU1Kedtkyews2FwvHDVQN87OoI7ntiObz+zG3/e2Ys/7+wt2GcThF7RpeAXIg//Iz9+TelZG4jENHaFySAifCPG5AjRlmZaJgBF8MVFxGHVetYpI3yLEb6QNv8dkKLvu5/6AG8fGkr78zPl7UNS07KvX7kcNrNUz9/jD2s6ctnMBkRiXInS1Tz7QS/ufuqDKT8jGovjmp++kduBT0G/nGXU5w3itv/Zhtv+Z1vBPpsg9IouBT/fHj7nHHt6E7ZCcmtBZdJWJdKZTDJK2T1x5Q7CbjZqPPxUefiVdjOicT7B1hkcD+Pxd47h849tSfvzM+WDbg/qXBbMq3Uo1s2wvGBMIMYfShHl/7/fbsPj7xybsifvsZGAUh66EIzIlpTRoMuvOEEUhbL8a0juLDWS9FxtYwgcGQm+AcFITLFoJMFP/KrNpolZOqJxilgPIBDNR9Q193ONxx9BncsKxhisJmlydsQf0Ub48u9kLBjFawdSt548MTq5oPeMJDKQ1NVF88VxeSzHPYnPTb57IohyoywF/+CAT/NcWDuCRB5+4teTSYQv/HhhE9nMxmk9fCH4X3r8PWxW2TeH5bGqo+1cI1XHNCljDUVjGPGljvD/+udv4TO/eged/eN45oPjaL/rWWWf7hH/pJ/R40m8lu6q5WwQQr9bdSfno5XCRJlTloJ/ZChZ8KUIX4ieyJO3ayL89KNSUSZBLKSyJwm+mCNQoy6/vKNHmruIxOL43VapH7zDmplIDo2HMBaMTL8jpKhd3EFYTQb4wzEc9wTQUm1X9hF9e48OS8I97Avj357frzlOt2fydQQ9IwEYGHDTOe0FKcLWl8I+GirgpDFB6BFdCn6+J22Ts2FEJk5jhQ0ANCttBZlYOlaTEW8dGlImWm0WA2yq94tetmoqVYJ/oG8cALC3dwzvHhnRjDFdrv7pG1j5zeexo1v7OxwcD+HBlzs19sZ4KAqXvCbBZjaia8iHaJxjXo1D2Sd50tobiGi6YQFA98jkgt/tCaCxwoYKu1RmOZ5i8jeXePwTf1/JVUAJotzQpeDne9JWTC5es6oZ9W6rUiiswS2tKrWoVtoK7NMsvFIjhPD53X0wMOl4lVM0UAG0gr+/X+oyJdYALGpwKY/TYTQQQY8cbX9nwx7Na5t29+EHG/dpxHosmLB0rCYD9ssXnDa14CedvzfF3cOJUa1f/pf9A4qwj/ojqHFa4CxQO0dPYKK4X//w23n9TILQO7oU/HwjKll+/9pT0OC2KqmG9bLgi9hTHZVnEuH3qKwNu9kIxhjmVNqmfI/a0hGFy4QlM6/GAY8/ktak49B4CKfe+zwAoMphxv6+MQz7wvjZKwcRicWVLCD1xPVYMAK3sHRU8xZzVYKv3g4AR4a0fn2D24o+b8IyeXlfPz77yDt46NVD0meEJNvIIX+OL8+2zqgqwhcXrnAsnrbNRRCzkbIUfBHhmw0GzQRqx7xqAMB+uY+rzTSzSduvXbFswvsa3FMLvluVhTMwFgLnXImi22odiMa5siZgKv64/bjy+PzF9RjyhfHl323H9/68F5t29yl174Xgh6NxhKLxxKStbN0YGDQXqeQIf+8J7WrZpU1ujW8+JNsn3/vzXrywuw8+WfBFhK/usJVrYnEObzCKS5ZJC8mODvvxsxtOA6D9/RBEuVGegh+Lw2xkMBiYppDZ5Svn4M5Ll+AbV60AkCh1DGS28OoL5y1AY4VcdEwWSmOKiVo1BgNDU4UNbpsJoWgcY6EovAFJnJsrpclTXxqC/7IqrfNDS+oBAK/ul9Io179/XLFShOCLY7qSIvw6lxUm1cUw+fz39I5pnrfVODSCP6CaIP3Cr7eizxuEy2ZSJr/zGeGLKP7shXU4uaUC3/34SsyrlTqYff2PO9HZPzbV2wli1lKegh+NKz69OsK3W4z4u4sX49LljQCAlqpElophGsFORnj+mXj/b//zxbjvmpMBAP1eKcuGMaBSzoePRKe2dN7sHMTrnYNor3XgC2vn49S5VcprDosRL+7px4lRSYhFcTRRHE5M2tbLDVzEBLYg2dIR2TqCpgobRvwRZbFZr0cS/5vOaQcgTZi6rCY45WwjfziG/X1jWPu9l6bM358J6hpGz3zpPFy3pg3zahP21Afd+e2kRhB6pXwF3zRR8JOj2GTRywQR2autoEUNrmnfJyaOB8ZC8AajcFtNSuG2cGzqapWf+uVmAMANZ87D165cjoX1TuW1b1y1HOFYHC/t7QMADMuWy1hIEkdh6YhevskXOPWkshij2oZqqEiMG5AqkJ7U5MbHT2tR9nFZVRF+KIpH3+hC90gA69/Prc3ikSe41WN2Wk343LntABJprwRRbpDgy4JvMRkmiNx0NsxUCMFXe98b/u487Pv2uinfVycEfzwEbzACt82s3I1EphB8dW67mGxljOG2Dy2E22rCOQvrACQWmYkIX9ScEb1422XBDyRZLqJ1I5Dw9ted3KRsq5Hr5os1Dcc9ATRV2jC3OhFZJ0f4Yj1CzxT5+zNBZDSpVwoDwDeuWoFVc6uw7wRZOkR5UpYNUMKxhOBbTBMXWeUCsUpXU0MnjYqbNbLwjvjC8AaiqLCblYvSVIJ/XLZQ6t1WXLC0Xtn+T+uW4p/WLZ1QTkI875LTM4XQi4tFqtXAb919EXb1eNE7GoDFZMDN581HQ4UVZqNBqZs/4o8gHufoGvJhzfwajei6bCY4VRH+EdkW2nU8u4i7fyyIrV0juHzlHADAwX4prbSlyjFh3wa3dUKGEUGUC7oUfM75egDrOzo6bsnH8cOxhIdvMkwso6Dmoc+cnlEOvCDh4Wd2E1VlF8IZliN8k1KueWrBl6LkBz91muYiIxZ5Ja8UFvWDjgz54bKalAh/fq0Tt31oIa49vXXCZ8yptGOOPIH8mbPbAQAnNVUAkEosi+Oe8AbhD8ewqMEFxhhqnBYM+8JwWk1Keqs/HMMh+T0z+f2qufmxrdjRM4od37wMz+08gW89sxsOixFNKVJha10WbDs6ktZxP+j2YEmje0KGEkGUKmVs6Uh/xOYUi6zUfHhFEz7RMTfjzxAiUWXPrAaOyWhAhc2EEV8YY8EoKmxmJZMoLE/a3vnkdjz86kHN+3rlRU+T5fvbzFJjFoGwdLqGfGivcygXBoOB4a6PnJTWfIOaGrnuzog/rIj/wnrpGIvkfx0WI5yy7z/sCytWji8k1e5JVYkzHQ7IWTeD42Glkctkv4dapxXDvvCElb7xOMdzO3rx3I5enPvdl7DvxBiu/ukbuP/ZPSmPQxClSPkKviyiwtLJdRQnjlfjyrzoWY3TgmF/BN5ABBW2xKRtJBZHLM7x1LYefGfDXs179p0Yh8VoSBnVAonGLAIxaXugbxzz6zIT91RU2M1gTLJ0hE20QJ40Xt4s3QX4QlFYTQYYmFTUjHPJZ/cGI1h93yZ8+YntWY1hYCykFGZ76DOnp9yn1mVBnCcmdgV/eK8Hf/ubbfjb32xDjyeAh+QLqliTQRCzgfIV/KRJ21w31Y7KEWStM3PBr3Za4PGHMRaMTPDwu5IKvwFSGYONu07gvMV1Kb13gVrwfXKBtB5PAKe0ZF/CwmhgqLKbMeILo88bgtHAlIneOy9bgs+ePQ9XntIMxhicFhN2yZkyK1sqldTQ53aeyGoMg+MhHB3246OrmrGowZ1yHzFHMuzTFlJLrtX/1LYeAEBDFplaBKE3ylPwYxMFP5OFVenglxc01cxE8B0WDI6HMRaKosJm0gj+f76csHIOD/rw4Mud6JGF+wK5ReFkiIuasHb+Ii/IOjkHgi/GPeIPY2AshFqnRclyqrCZce81Jyt2jsNqVOrVr2jWfvZMataLtxz3BNA7GtCUhEhGXISSC6lN9rl+6olLzCJ0OWmbb8LROCrkvHOTKIWc4whfrCSdqeBv6RoG54DblojwPf4I/m9bt7LfjY9sxrHhgLJArL12cqEDEvWAGt02nPAG8ewHvTAbGVa25kbwGyqsOO4JoNJuVvLyU2GQrzgXn9Sg5PQLukcCGBgPoa3GoUkFnQzOE20X3z0ygjiHsqo2FYkIXyv44i4DkKyoQ3IfgkEfVdgkZg/lGeGrLB2RrZPrjlKiSNl0VTJTUeM0KwJUYTcpYxS2w9pFUk79sWFp0vOtg1IZZvXK4FQIS6dVrnP/eucgzl5Yl7NzX1jvQmf/OPrHQsqK3VT0ytH9dWvalLx8wa7jXnz8P9/EtT97c8L7/rJ/AGd950XNmgNvIKrYZy/ukcpKrJzijqVWnlNR18bfdXxUKfIGAB9dJS0Wc1iMVEOfmFWUp+DHElk64k6+bQobYEafIRdoy6RxiqBK1WnKbTMrLRFFrZqlTVp/+o2DgwCA5mkEX1yE1BH9OQtrMx7fZCxucMEbjGLXcW9a0fnKlkrl9+OwGGFgwPZjHgBAV4pc+e//eS9OeIOaOj4D4wnvPRyLw242TplhJLKJ1JbOv27cp9nn1vMXYNPfn48bzmxTisARxGygPAVfVUtHRJut1VOLZab8x/Wr8emz2jJObwS0NlCFytIRNWfmJo21eySAerd12kwjscr2jPYaZdt0NlAmqCdK0ykn3VhhVSL8KrsZ7XVOvKVq75iMWMSlbqUoVgqLkhCntFZOuULaZDSgymHWWDrJjc5tZiMWN7pR5bDIrSqpNSIxOyiYh88YWwDgHgCVnPNrC/W5qQipLJ1jsni05FjwFze68e2PrpzRe6s1EX5i0lbUm5+TIpJvnMIzF3jk3Ht1IbHW6twJ/pr5NfjqZUtQ7bTg0mWNk+73yxs7MOQLgTGmRPhWsxGt1Q5s7Rqe9H3CehLNW+Jxjn75d7Kk0YXDgz4lBXQqap0WDKmydLqGfFjS6MKPr1utrAQGEhZYKBKnxVfErCCtCJ8x9ghjrJ8xtjNp+zrG2D7GWCdj7K6pjsE5P8Q5vzmbweaKcDSm5OFfs6oZAHByc366a82EalU5Arct4eELSyfVoqJ05gqE1632+ttyGOFbTAbcftFi3HDmvCnTGS9Z3ohPntEGICGqVpMBLqtRsZ1S4ZNr6AvBv+bBN/Dl30m5+x9bLa0MvvrU5mnHWeu0KlZNLM7RNejDxcsasWxOheb3YS9Qdy6CKBTpRviPAfgpgF+LDYwxI4AHAVwKoBvAFsbY0wCMAB5Iev/nOef90AmRGFci/BvOnIfrz2jLuPxxPlFbOi3VdhjlrJYhXxhmI0vpj6cj+AvrnTg44EOl3Yx6txUDYyFU2DKfVM4l4iJkNRs10XUqhmQbRlhT6qqXH17RiL33rUsrEq91WXBArrcz4g8jGucpL6Ki3AYJPjFbSEvwOeevMsbakzavAdDJOT8EAIyxJwBcwzl/AMCVMx0QY+xWALcCQFtb20wPMyXqPHwg81r3+UbdeMWq6kAV55Ldk6oMRDqC/+TfnI0+r2SlPPt3azE4VvwJyeVzKnBySwW+edVy/GmablRisZQvFJ2QN88YS9t2qXVZ8PYhuS+AfBFR22gC8XsOTHHXQRClRDYefguAY6rn3QDOnGxnxlgtgPsBrGaM3S1fGCbAOX8YwMMA0NHRkfkqnGmIx6W87alWpBabqhTibTYaEIrGUeO0aISt0m7GaCCi6Yk7GbUuK2rlu4MGt23atouFwG4x4pkvnQcgkVYpCEZiyrlyzhVxHg9G4ZNF+Mz5Nfir0yYWepuKGqcVnkAEX/3f93FUzgZKtSJafHZwhjV+CEJvFGzSlnM+BOC2dPbNZ3lkYSGYdBbVqzEZDfjqZUtwrpxvD0jrBULROKodFqW2DiDZP6OBSNGtmVzgsmm/jh5/BE2VkugO+cKIxKT/u/FQVJmA/vhpLfjEGZkVt6tzWcA58NS2bogaaqlqHokIP0gRPjFLyCbM7QGg/ktrlbdlDed8Pef81srK3E+kilWZJh1H+ABw+0WLsbqtWnkuSiTXuCwaC0qkP+r5ApYuzqQFYKKJOyClngLShPN4KKqUVK7MsBopkJgjURfMTLUiWkT4wsN//cAg2u96VinrTBClRjaqtwXAYsbYfMaYBcB1AJ7OxaAYY1cxxh4eHc19K7pIXFoQVWoCKUok1yR5zWKiM+feVxFwJa26feT1w0oOvMi9XzbHLUf4qbtapUOtc+Kkd0oPPylL54/bpXjmzYOTrxUgCD2Tblrm4wDeArCUMdbNGLuZcx4FcDuAjQD2AHiSc74rF4PKa4Qf07+lkwpRfyY5EhWiNIOaY7ojOUvniS3H8MNN+wEkIvylTW7E4lxJUZ1J6YraFPZNqjkdxdKJSEGCyNoZp4JqRImSbpbO9ZNs3wBgQ05HlGdEhG/UuaWTjG+S6pt/d/FixDnHJzP0sfVIqpo+vaNBvLinD+8cHkal3YwmOb9fXABmFuFrf4fuSWoJWZPSMsUaAWqRSJQquqyWmddJWznCN5dYhC+iyuoksZpX68B/3zxpclRJIeZVGiusyqri53b0Yv37UrrmiuYKZWL3h5v2o8ZpmVE10iqHBW6rCZevnIOn3z+On95wWsr9kidtxV2FaPBCEKWGLsPcQkzaTlVvRY/ccfESLKhzTmhWkk7NmlLBLYv5ZcublG1R1cxqa7UdLmsiov/KZUuUdQqZYDQwPPfl84WRogwAACAASURBVHDvNSuw5751+NCS+pT7qSdtDw6M441Oybs/LreTJIhSQ5cRfj4RjcD1nIefijsuWYw7Llk8YXuuG7cUk2VzKvCHL56Dk1sq8d9vH5nweoPbpimnLHrmzoR0agiZjQaYDAzBSAxPbpGWnLhtJvSOBsE5V/oAE0SpoEvVy2eWTiItc3b8septlXC2rG6rnvRiHOccC1T9d3Nd0joVdrMRgUgMg+NhNFfa8JVLlyAcjStlHvTOf7x4AN98Oie5FMQsQJeCn09LJ1KiWToEsGpulaZJe1MB+s3aLEYEwjEM+UKodVmVSqW9nuA079QH/7ZpPx57s0vTNIYoX3Qp+PlEifANZXfqJcv8OideuPN8XHu6VELh46e1YFGDqyB3N06LEb5wDEPjYdS6LGiulAS/1Hz8Nztp7QChUw8/n1k6ibTM0o7wf3ljB44Ml0d6oNNq1DRX+eEnVhXssyvsZowFIxgaD2FJoxuNldKirX5vaUT4gnL5rhBTo8swtxBZOuYSj/AvWd6Im9fOL/Yw8sYbd12ENXJnrunKJucTt80EbyCCIV8YdS4LahwWMAYMlEDrQ5GgAAADY9Sbl9Cp4OcT8UdQammZ5UZLlR2LG6UJ2lw3mM8Et9WME6NBhKJx1LosMBkNqHFYMKiz5uaBcAwnff05/HbzUWWbKD8BAP1jpXVHQuSHshN8JcIvcUunHBB9asOqSLXQuG0mHJd7CYvm8nUuKwZ1FjH3eAIIRuL45z/sULaJiqIARfiEhC4FP59pmWKlrd6rZRJS3RwA6Bkp3gSpW1V2WhRYq3NbMKCzCF/dlF00bBHbnBYjCT4BQKeCn9+0zNKsllmOnNQkNSQf8RfPL6+wJ+wkUaitzmXVnaUzpBrPsF/bCnJxoxsHB8Z1N2ai8OhS8PPJbFt4NZupd1vxhbXz8YsbO4o2BnWELwq1SZaOviZtB1UR/oj8WFg6N53Tjmic4ycvHijK2Aj9oMu0zHwSKYGOV0SCr125vKif77ZNjPCr7GYEIjGEo9reyMVEHeGLVcAi0v/wiiasXVSHzYeGizI2Qj/o49taQGJKA5SyO3ViBlSkEPxKOdIXXbf0wNB4qgg/ApvZALvFiDXtNdjXN4ZRv37GTBSeslM9UVqB0jKJdGircSqPRfVMIfx6EPz/e7cbJ339OZzwBlEnN3ZRInxfWJloXiJPgB+lBVhljS4FvxDF00qtWiZRHJY0TqzIWaEjwb9/wx4EI3HsPu7FogYXjAam8fDVqaQAaOK2zNGl6uUzSycqsnRo0pZIg1TpuyLC9+pA8EXryx5PAHMq7ah2mDVZOjVOaawNbknwKT2zvCm7SdsoTdoSGbL+9rXwqapNKoIfLL7gq7/HDRVWOK0mpR3miC+MZc1SaquI8PW2foAoLOUn+LTwisiQla3aO009ePiRWBxGxjRzUY1uG+xmI3yhKH74/D4cGvRh7eI6AFKze6fFSJZOmVN+gk8RPpElFXJufjEzXpZ+7TlcvKwR6mSzxgobHBYj3jvqwQt7+gEAy+dUKK/Xua0YLIGib0T+KLswN0orbYkssZgMsJuNRYvwezwBxDmwaXefZntrtR0Oi0nTjevUuVXKYz3WACIKS9kJfqREm5gT+qLCbsJ4qPBdpCKxOD77yDsAAAMDxoKJMaxsqYQ9qan94oZEllG9DktCEIVFl4Kf37TMOEwGRg2oiaxwWU0YK4Lgv3fUg87+cQDSXMJoQFpc9dBnTofBwOCQBd9mNmD/tz+imavKVdG3eJxjf99Y1schCo8uBT+/aZmconsia1w2M8aDuRH8NzsHcSzNBVF/2d8Po4Hhc+e2Y8QfAefAVy5dig+vaAIARfDnVjsmlH2oc1nh8Uc0jVEyJRqL4z9e6sRlP3oVW7qoVEOpoUvBzyfROKdFV0TWuK0mjM0gLZNzjrhsK4rnn/rlZnz4319N6/2HBnxor3VgQV1iBbC63o/dLD1uTNHgXaRmDmUxcfupX2zGj17YDwB4o3NwxschikPZKV80FqdFV0TWuKxaDz8e53j9wGDK6PnBlzvxt//zLgDgxkfewSn3Pq+8Jmwhv1zDfjqGfGHUuqyKeAOAU9URTET4osyCmmxX2/aOBvCOKqrf2jUyo+MQxaPsBH/viTE0pYh+CCITXDaTxtK5f8MefPpXm7FhR++EfX+wcR+e23kCAPDagUGMh6LgXIryT4xm1npwaDwk9dZ1JgTdpY7wZcFXl3UW1LuzW3zlSUpD7Rryzeg4RPEoK8EfDUSw9cgILl7WUOyhECWO26adtH3mg+MApIDigQ178Mjrhye8J6aycvrl9MjeDAV/2BdGjdOCarXgp+j567AaJ2yrld8zPENLR9zRrGypxPI5FegeCeCVff1pvfc3m4/gV68fVi50R4f82HeCJn4LTVkJftegD7E4x+q51cUeClHiuGVLh3MObzCCPq8k4PtOjOGhVw/hW8/snvAedY/ZQwNSdNzrSbRvDEamtnWisThG/BHUOq2osicieLXg++USEA7zxIuAqJzpmeH6ATFncd9HT8YNZ7UBAG56dEta773nDztx3zO7sf2YBwBw/g9eTnvegsgdZSX4Ime5wj7xdpcgMsFlM4FzwBeOYb8cqbptpimjVvWCqD29XgDAcZXgT2fviJaFdS6LUpMfSBZ86aLhTBHhu20mMKa98GSC+PtxWU1oqbIr26e7UKlR994FtHc9RP4pK8EfD0l/MKlugQkiE1xWSXDHg1Fs2t0HAwMuW96EHo+24bqwMACgeySRevmtZ3Zjw45e7FVdIKYrxjbkk+4iapxWWE0JQVd/nzvm1QAAVrdVIRmDgaHSbp7gxaeLsHQqbFrB7/EEpsxYiqomspMnp3tHi9egvhwpK8H3yhGKOo2NIGaCKDe8/ZgHv918FJevnINlc9wT9guool+xYOqvT28FADy9/Tj2nPAqGTW+0NSRctegdMFoqbZrtquzdK44ZQ623HMJTpeFP5lqh2XGlo6YpHbZTFjc6MbNa+cDAC7+t79gzf0vTv4+1VyHP6xdu3B0iBqyFJKCCT5j7KOMsV8wxn7HGLusUJ+rRrF0UmQwEEQmXLC0HvNqHfjJiwcwFopi1dwqTaqkQB1NC8G/eFkDzl9Sj70nvDg2HFCict80K3d393phYMDSRu2FJXmBlcjGSYUU4c/c0jEwwC53/jp/Sb3ymvrCNhqI4INuj7LewBtInFfyRY06cBWWtASfMfYIY6yfMbYzafs6xtg+xlgnY+yuqY7BOf8j5/wWALcB+OTMhzxzRISSyt8kiEwwGQ246KQG7Ja9+DqXFbVJue+xONcIvrB7qhwWtFTZ0CVHtx3tUhKBLzyN4B/3YkG9a0K9nEyodmRn6bisJqUsSWXSXFhAtmu+/+e9uPqnb+CxN7twoG9MU2QuEIlp1ioUox5ROZOut/EYgJ8C+LXYwBgzAngQwKUAugFsYYw9DcAI4IGk93+ecy7yt74mv6/gjAUjcFiMVAufyAmrkipRqnPjAUnMjqhy1UW3KbfNhObKhC3T0S5F+NMtvtrT68Xp87LLMKtyWNA5MD6j944Fo5r8/ooka3TZv/wZ//DhpcrEtchU+uIFC5V9fKGopuBbKDrzMg9E5qQl+JzzVxlj7Umb1wDo5JwfAgDG2BMAruGcPwDgyuRjMCks+C6A5zjn2yb7LMbYrQBuBYC2trZ0hpc20heW/HsiN6jTe+vcFtQ4Jgr+S3v7UWEzIRiNK7Xo3VYzmlWTnvNrpTIJU1k6o/4IejwBfObsecq2F+780ISsl+mYW23H0+8HcWhgHAvqJ/brnYrxUETz95Mc4QPSIrNkS2nrkcSKXH84pmkNGUhzhTGRG7IJdVsAHFM975a3TcaXAFwC4FrG2G2T7cQ5f5hz3sE576ivr59stxkhbkkJIhfMrbErUX3KCD8YxVuHhrB2cR0qbCZFnJ1WI9bMr8Hqtir8/SVLlEVSU0X4wjpSNzRZ1ODCmvmpJ2cn4zNnt8PIGH635dj0OyfhDWgDpsnSm5P75u7oTlS99YejmmykQAYpnUT2FMzb4Jz/hHN+Ouf8Ns75z6faN1/lkb3BSMol5wQxExhjWDW3CgYmZb+YjAasv30tHvzUaQCklMPukQBWNFdq0iidVhPm1jjwhy+eizsuWQyz0QCLyTClhy/KEZ/UNDETKBPq3VYsqHcqE8iZ4AlEUGlPXNTMRgPqXBasbqvCZ1V3HskIUW9wW+ELxyZ4+kThyCbc7QEwV/W8Vd6WNZzz9QDWd3R03JKL4wnI0iFyzfVr2tBabVdKbq9srURYnpTcJlsZSxvdykSr2chgNU2Ms5wWI/xTpGV2j/hhMxumzMBJlwX1Try6fxAjvrCmRMN0jPrDWNFcodn22j9eBJtZOp+GCht+sHEfPn5aC3Z0j+JA/ziaKmw44Q3CwKSLjT8UxRF5stpoYAiSpVNQsonwtwBYzBibzxizALgOwNO5GVZ+GA+R4BO55dLljfjWNSdrtglr542DQwCApU1uJZXRqcpyUeOwmKaM8Hs8ATRX2XPSuGdOpR3joShueiy9sgiC0UBEU9IBkIq1MSY1FLp57Xz89gtn4t/++lT807qT4LKacO4iqYl6hd0Ml9UEfziGgwPjsJuNmF/npAi/wKSblvk4gLcALGWMdTPGbuacRwHcDmAjgD0AnuSc78rFoPJl6YwFI3BbydIh8otIz3z3yAjq3Va0VtuVKHiyOSSndeoIv2ckoFndmg0XnSQVD3xfrmuTDuFoHL5wLOVErcBmNuKcRXVgjOGS5Y3Yee+HlUViTosJDotRFnwfFjY44bQYSfALTFqCzzm/nnM+h3Nu5py3cs5/JW/fwDlfwjlfyDm/P1eDylfHK7J0iELgtppgkVN/z15QC8YYbHKEP5ngTxfhd48E0FqdG8E/d1EdvnTRIhiYJOTpIHz3KkdmAVO1an+HVTrHrkEf5te5YDMbKUunwOgyIT0fEX4szuEPxzS1wwkiHzDGFB9f5M2rLZ1UOK3GSdMyo7E4hnzhlF2sZkp7rRNxDhwbSW+l62hAyjCqdKTv+QOJCp1xzqWIPhzDiD+MWqcFdosxo8JrRPboUvDzEeGPK3V0yNIhCscSuQyCmLSdLMJ3W82aBUlqRDmCXH5359dLuf8H0mxGrkT4GVaaFVU945zDYZGaxoyHoqiwm2E3k6VTaHQp+PlA5P6SpUMUksWN0uImm2kawbeZJhX8MaXKa+5KgpzcXIkKmwmbdqfXwGTEJ41hKg8/FYkIX2q/OBaKgnNplS4JfuHRpeDnw9IRNTvctPCKKADC6xZdpqxK6mLqtEq3zTxpiWER4btymHBgMRlw4UkNePXAQFr7izpAc6oys5XEHUE8zjV2VoXdDJvFiECYSisUEl0Kfj4snTGydIgC8twd5+GZL61V0iiH5LIKSxpTL5yqsJvgC8cQjcXxlSffx1eefF95TfRxyHXRv6VNbgyMhVLOHYz4wprmJN0jflhNBtSnqAg6FeKOoN5tVRqsA1LFWrvZiBBF+AVFl4KfD8QfDVk6RCGYU2nHyS2JgEVMji5qSF2/RgQi46EoPuj24J2uIbx5cBD9Y0GMKx5+br+7bTUOABNLFHuDEay+bxO+v3FvYvzDUpZQpusAqp0W3P+xk/Grm87QCr7dBJvZAH8kpmkSQ+QXXQp+PiydMVXzBoIoNKe1Sdk6iyYpWCbEfCwYxcB4CD0jAXzqF5ux5v4XFatnsgyfmTKvRpq43dmj/TvzyH79U9sSC+e7PX7MlS8QmXLDmfPQUmWHw6KydGxm1DitiMW50rqRyD+6FPx8WDoi39eRRS1xgpgp/3z5Mrz2jxdOWspAlBoe9oXh8UegbvW6aXcfgNy35hQR/j/8/gNNQTPxWF23vtcTxJzK7NYBqC2pSrt50jsMIn/oUvDzgcj3tZlI8InCYzEZpoyQRRe2w4O+Ca/9aftxALkX/EqHWamNM6iqcCkapETkRVnxOMeIP6y0YpwpdrM2wifBLzxlI/ii0YLIliAIPSE8/EMpBF+Qa0sHAP7+kiUAoEkJFTn3kZjcojAo3XFUZ7joKhl1hO+ymTC3RrpjOEaCXzDKRv2CEVnwKcIndIjIZplqIZQ5D53a1HMHAo+8qlasFhZ1/JPr/WeK2sM3GhgcFhMaK6wzKtVMzAxdCn4+Jm1D0RjMRqaUsSUIPdFcZYPZyPCuXFI5uX1gvhB3Fuo1AOp69cLOAZBRKeVUpJo/W9lSlVERNyI7dJmyko96+MFInPx7QreYjAa01zpxQI52VzRXYkvXMDbccR4C01SpzIZUEb5a8MdCUQzLWTvJLRxn+lmfPivRunR1WxVe2NOHofEQajPM8ScyR5cRfj4IRmPk3xO6Rp2j/+mz5uHGs9uxpNGNU+dWob3OmZfPFJPF3mAEkVgc3SN+jKrSJL2BCEZ8IsLP7qLjtpnx1t0X4d6rE/0DLlzaAAMD/v2FA1kdezpe2dePv+xPb1XxbKZsFDAUiZN/T+iaeXIzc6OB4fKVTfiXq5bn/TNdqgj/py91Yu33XsZ2lcUyGohg2J8bDx+QFqSpbdXlzRU4f0m9YmXli5se3YLPPvJO1sfpGvThxkfeUe6CRv0RePyZNZIvJmUj+BThE3pH1I5nQE46W6WD0cDgskqF20Tf3L0nxhRxFxG+1WRQSjznmkq7Gf4pegFkSzyeu5W8Dzy3B6/uH8DLe6Wic1/+3XtY9a1NJTPxrEsFzMukLXn4hM4RPn00hwKVDm6bCd5gROnUBQCr5lYBkCN8Xxg1TkveLkJS85f81dTpGwtmtH9skt8/51wpwjg4Lq1beHmfZBO9c3g4ixEWDl0Kfj5W2oaiMaXNHEHokUy7SeUKqTRzRFPiQAi+NxjBiD+cdQ7+VEgN3PMX4R8emHxtQzL3/GEHTv7Gxgn1fbzBCM64/wW80Sn1KT44IEX0TjnzSFwA9E7ZKCB5+ITeqbTnT1SnQirNHMWoP4JKuxmt1Xal7606ws8XDqsJ/kgsp9aLGnWEry4XkYrfbD6KQCSGzUkR++7jXgyOh9FSZYfFaEBn/ziisbhyZ0KCrzOCFOETOidfqZfTIZqvjPjD6JhXjdf/6SKsaK6A0cDQMxLAiD+SdQ7+VDgtRnAu/Y3mA3XN/cmazAhEo/g3Dw5ptot02d//7dn4q9Nb0dk/rrkjIsHXGcFIjCJ8QtcUz9KRmq94/BGlJSFjDLE4x3+9dQSHB32oyePYHHLJCNHoJdeoJ4THghH4QtFJm6eLwnE9IwHN9s6+MbisJjRV2LCowYURfwQH+hOrogfHSiNTp2wEPxSNU4RP6JpiRfgVcoTvSfLqL1xarzy25ilDB0j44LnO1PnZKwfxnQ17NI3S3zw4hBXf2Iiv/v79Cfv7w1HlDqA7qbl715Af8+ucYIwp6yXERG2dy0IRvt6gCJ/QO8Uq3e22mTHkC8MXjimpoQDw6OfW4IlbzwKQ34uRqLGT6wj/e3/ei4dfPaSpxvnvL+wHALydZNkAQL83IdrdSRH+sC9RLTRZ8Bc3uDFQIoKvy9IK+YAifELvMMbwL1cux+nzqgv6uepOWsk1789aUIv1t69VmrHnA1FFc7Kevtny9PvHlcd9sqiL6rkAcHTIj2g8jn65RPTqtiq8d9SDP23vwTWrWgBIgi9+B82VNjgsRmyVF4u11znw9uEhxOMcBp3X6tKlAmabh3/HE+/hpke1q+qCkVheb0sJIhd8fu18nCqnRBYKdaG2ttqJNftXtlbClse/HRHhf/Lhtyf11jNF3dAlGImjymHGx1a34J7Ll+GOixdjPBTFWDCCI0M+nP+Dl/FXP3sTfV4pm+eW8xagscKKf35qh9INbNgXVhrSM8awsN6FsHzRmFvjAOfTTwjrAV0KfrZ5+JwDh1S5t5xzKcI36fJ0CaKoiIqZADC3emZtDLNBXSf/j9t7ptgzfZInXavsZvzok6twy/kLsKBeKmFxYjSIHo+034g/otg4axfX4TdfOBO+cAz/8PsPEAjHEIjENJlKwtZRN3b35ukOJZfMSgWcU2XDidGgktcbjsXBeX4nngiiVFFbOg3uwlesVF9w3j400VufCULwz15QCwCaO5RmOfVyX98Y/Kp5g509o7CbjXBbTVjU4MaNZ89D94hfKQ+trhY6T74TclpNyvyGusqoXpmVgt9caUc4FseQXOUv0fxkVp4uQWRFlSxkzZW2onjQLVV2PH7LWThnYa2ygjVbhPieNMcNAJpMnZObK9Fe68C/btwHnyozaPsxDxorrEoJicYKG8aCUfSOShcP9eKzZnmuIxSJoUIWfC8JfnGYU2kDAOU/KiQv6MinD0kQpcqquVX4yfWr8fydHyraGM5eWIsljW4cGvBNKGswE8QE8IJ6yXoZV0XydosRV5wyB90jAaU2DgD0jgbRUGFTnousnH0npIuQWvCbZI3xqXoVUIRfJMQt23GPNAkTogifICbFaGC4+tTmnDdJz5SFDS74wzGc8GZW7CwVQsgXyH0ExkNaMa52WBCNc/SNaj+rSSX49bK9te2olI3TqprfEEElkEhZJQ+/SIiqf8J7owifIPRPS5UkoidGsxN8fziK97tHYTEZlFIJwtYViAVmYqJWlH5urEjMYdS7pPFs7RqGzWzQzG80qQS/giL84iL+84RvRx4+QeifWqckqEPj2ZUp+H+/2YZNu/tgNxuVKD0ZYc8cG/HDZjagtVq6MDSmiPC7hvyYV+PUzG+IieYLltbDaTHCZGAk+MVCRPIBWfApwicI/SNEeNiXneC/IrcyDEZicMo21eIG7cIxkWLZPRKQauTIEbvaw29wW5UxtddNTFd992uX4OefPh2MMdS6LFnX0xkLRnD5j1/Df77SmdVxpqJgph1jbBmAOwDUAXiRc/6zfH2WiOSDkThicY6H/nJIs50gCP0hrNihGQp+JBbHdzbsgZjzFatp//zl89Dotmn2FSUkekeDmFfrUDz5RtUdgcHAUOu0YNgXxtpFdSnGq7J/3Nasyyt0Dfqxu9eL3b1efPGCRVkdazLSUkDG2COMsX7G2M6k7esYY/sYY52MsbumOgbnfA/n/DYAnwBw7syHnNZ4YTUZEIrE8KftPXh+dx8AivAJQs84LCbYzAY89OpB+GbQEOWlvf149I2uCdtPaqqYUN5Z/dxhMSklJdTePADcc8UyXLi0Hp84Y+6Un13vsqI/w85aABCOxhXredCXuGCE8lQqOt2Q9zEA69QbGGNGAA8C+AiA5QCuZ4wtZ4ytZIw9k/TTIL/nagDPAtiQszOYBLvFiEAkpqlrTT1tCULfBCNxePwRPPhy5rbGkaH0O1u5rSZY5Dt+l9WIC09qwKXLG5UMP8EFSxvw6OfWTFt4sd5txcBY5hH+1T99HafdtwkAMKyau8hXqei0LB3O+auMsfakzWsAdHLODwEAY+wJANdwzh8AcOUkx3kawNOMsWcB/Hamg04Hm8mIYCSm+RJQT1uCKA2mmrj1haIwGtiEO/bkCpdTwRhDa5UdhwZ9cFhMWDW3Cr+4sWPG421w2zA4HkYszmHMYPHa3hNSTf1YnGNIFeH7QtG8dBnLJuRtAXBM9bxb3pYSxtgFjLGfMMYewhQRPmPsVsbYVsbY1oGBgRkPzmY2IBiJa66UFOEThL75/rWnAAAOD04erZ/9wIu48j9eBwAcG/bjM7/ajIGxEI4Na2vY33v1iik/q0XOzBEZOtlQ77YiFucZTThHVe0Wjw77NXMX43nq8VuwSVvO+SsAXkljv4cBPAwAHR0dM15yZzNLlo66qQJF+AShbz7RMRfbj3nw7Ae9KV+PxOLwBqPwBsfx1sEh/GDjXmw76sGbBwdxZNiP9loHuob8uPKUOfjsOe1pfeaK5pkVaVQjvP8To8FJU0GTEQtDAWBvrzfJ0tGf4PcAUM9ktMrbsoYxdhWAqxYtmvlMtc0sWTo+VblVmrQlCP3TUmXHaCCCjbtO4MMrmjSvHRlKRPF3Prld6TT1QfcoDg34cM/ly1DntuCipY3Tfo6ombNQrp6Z7ZgBqVPWytb0LiAHBxN1gw4N+goS4WfjcWwBsJgxNp8xZgFwHYCnczGobMsjA5KlE4rE4Vf94igtkyD0j1j89Df//S7e7BzUvNYp95Fdt6IJvaNBRGKSCfCr1w/DwICrVzXjY6tbld68U3HfNStw0zntOWk4I8pKi3LLao4N+3HOAy/i6JAf/d6gIuY7ukfBmNTi8dCAD0eGfJgvl4LI16RtummZjwN4C8BSxlg3Y+xmznkUwO0ANgLYA+BJzvmuXAwq2wYogLTa1h+JaiJ8vXejIQhCW6L5tSTBFxH+R1dPnC686tRmzUrZ6ZhX68Q3r14BkzH7QLDCboLLalImjv3hKO5dvwuj/gj+d+sxHB8N4vfvHsOa77yIjm9vAucc7x0dwaJ6F1a2VqKzfwxHhvw4Rb47SK79kyvSzdK5fpLtG5CHFEvO+XoA6zs6Om6Z6TFsZiNGfPpf6kwQhBa1aL8nFy4T9I+F4LAYceb8mgnv+0TH1Lny+YQxhtZquyL4//P2ETz6Rhcq7WaE5bsQszGxIPSpbT3YfHgYV53SDIMBePwdKf/llNYq/Gn7cU11z1yiS48jFxG+zWxU0pzuu2YFDn3n8lwNjyCIPKIuYDaYlJ7Z5w2iwW1FtdOiqVgJoOC9gJORBN+P0UAED796GIDUfU+0QlS7DV/53/fhD8fwkZVNOGdhYhXvKrm9pR4nbfNGbiJ8g1I0zWUzkZ1DECWCKDcMAENJ5Qr6x0JokMskLJ9Tgd7RIO68dAmcVlPRkzJaquzYfHgYT245pkwm948FlRIPB/rGNPt3zKvG2kV1MDCGNw8OosFtw2ltVbCaDOUl+LkgrqqGKpokEwShfxhj2PT35+OJLcfwq9cPIxqLKz77wFgIK5orAAArmivwxyFejgAACBpJREFU4t5+fPqseXlZpJQprdUOjAWjOKTKvukeCSAuF/fZJwv+58+dj6VNLnzyjDZlvwc+fory2GY2KheJXKNLJcxFWuYiVXU8Jwk+QZQUixvdaJczVob9YSWq7/cGceHSBgDAjee0Y2GDSxdiDyQWcm3tGkFbjQMrWyux/agHMbm3tvD3rz29Fcvli1YqzEZD3gRflx5+LtIyv3DefOXW0GGl/HuCKDXqZCH/2INvgnOO7hE/fOEY2mokYa1zWXHNqkkX9xccsWL3QP84miptOK2tGj2ewIQOXnXuqS9QVpNB8f1zjS4FPxcwxnDNqmYAgCUHaVcEQRQWkUvf4wlg13EvXjsgpWiuXTyxVLEeUBdea6qw4cKl9crzK06ZozwW3bYmw2xkiMTKyNLJFV+/cjkuOqkBJ7dkv3SaIIjCsmJOJdpqHDg67Mfv3+1G/1gQcyptWFjvmv7NRaDWaZHFmqOp0ob5dU7cd80KnLuoDo+92QVAit7N0wSgljxG+LoU/Fx4+IDkhV0g+30EQZQWlQ4zXv3HC3Hnk9vx23eOIhyN4xMdrUpJBL3BGFNW/i5ucIExhs+c3Q4ASuetdBrFm42GvEX4uvQ6cuHhEwQxO7jtQwuViPf8JfXT7K0PkuvpCKG3pFHexWIyIEyWDkEQ5ciSRjduOW8+onGOdUnF1PSG02KELxzDoiTbSQg+T6P+ryWPWTok+ARB6J57rlhe7CGkxdNfWoujQ/4J9XnSsXIEFpOh9OvhZ0KuPHyCIIhCsrDelXJS2ZmJ4BvLLC2TPHyCIGYTblv6gl92k7YEQRCziYwifFp4RRAEUbpk6uGL9M5cQ4JPEASRZxyW9Mu7lF0tnVzUwycIgtALoq7XJzpap91XqqWTnwYouszSyUU9fIIgCL3gtJqw51vr0uqrLcoz5ANdCj5BEMRsw56mrZPPlba6tHQIgiDKFbPRgFicK3X0cwkJPkEQhI4Q9XbykYtPgk8QBKEjRP+OfNg6JPgEQRA6QkT4+Vh8RYJPEAShI5QIv1wEn/LwCYIoV0RHrLLx8Kl4GkEQ5UpbrQNXrJwDqyn91bnpQnn4BEEQOuKM9hqc0V6Tl2PrMsInCIIgcg8JPkEQRJlAgk8QBFEmkOATBEGUCST4BEEQZQIJPkEQRJlAgk8QBFEmkOATBEGUCYzz/HRWyQWMsQEAR2b49joAgzkcjp6gcytdZvP50bnph3mc8/rkjboW/GxgjG3lnHcUexz5gM6tdJnN50fnpn/I0iEIgigTSPAJgiDKhNks+A8XewB5hM6tdJnN50fnpnNmrYdPEARBaJnNET5BEAShggSfIAiiTJiVgs8YW8cY28cY62SM3VXs8aQDY+wRxlg/Y2ynalsNY2wTY+yA/G+1vJ0xxn4in98HjLHTVO/5rLz/AcbYZ4txLskwxuYyxl5mjO1mjO1ijN0hby/582OM2Rhj7zDG3pfP7V55+3zG2Gb5HH7HGLPI263y80759XbVse6Wt+9jjH24OGc0EcaYkTH2HmPsGfn5rDg3xlgXY2wHY2w7Y2yrvK3kv5NTwjmfVT8AjAAOAlgAwALgfQDLiz2uNMZ9PoDTAOxUbfs+gLvkx3cB+J78+HIAzwFgAM4CsFneXgPgkPxvtfy4WgfnNgfAafJjN4D9AJbPhvOTx+iSH5sBbJbH/CSA6+TtPwfwt/LjLwL4ufz4OgC/kx8vl7+rVgDz5e+wsdj/d/LY7gTwWwDPyM9nxbkB6AJQl7St5L+TU55zsQeQh//EswFsVD2/G8DdxR5XmmNvTxL8fQDmyI/nANgnP34IwPXJ+wG4HsBDqu2a/fTyA+BPAC6dbecHwAFgG4AzIa3KNCV/JwFsBHC2/Ngk78eSv6fq/Yp8Tq0AXgRwEYBn5LHOlnNLJfiz6juZ/DMbLZ0WAMdUz7vlbaVII+e8V358AkCj/Hiyc9T9ucu3+ashRcKz4vxky2M7gH4AmyBFsB7OeVTeRT1O5Rzk10cB1EKn5wbg3wH8I4C4/LwWs+fcOIDnGWPvMsZulbfNiu/kZFAT8xKBc84ZYyWdQ8sYcwH4PwBf5px7GWPKa6V8fpzzGIBVjLEqAH8AcFKRh5QTGGNXAujnnL/LGLug2OPJA2s55z2MsQYAmxhje9UvlvJ3cjJmY4TfA2Cu6nmrvK0U6WOMzQEA+d9+eftk56jbc2eMmSGJ/W8450/Jm2fN+QEA59wD4GVINkcVY0wEVOpxKucgv14JYAj6PLdzAVzNGOsC8AQkW+fHmB3nBs55j/xvP6QL9RrMsu9kMrNR8LcAWCxnElggTR49XeQxzZSnAYhZ/89C8r7F9hvlzIGzAIzKt6EbAVzGGKuWswsuk7cVFSaF8r8CsIdz/kPVSyV/foyxejmyB2PMDmluYg8k4b9W3i353MQ5XwvgJS6Zv08DuE7OdJkPYDGAdwpzFqnhnN/NOW/lnLdD+jt6iXN+A2bBuTHGnIwxt3gM6bu0E7PgOzklxZ5EyMcPpBn1/ZC81HuKPZ40x/w4gF4AEUg+4M2Q/M8XARwA8AKAGnlfBuBB+fx2AOhQHefzADrln88V+7zkMa2F5Jd+AGC7/HP5bDg/AKcAeE8+t50A/kXevgCSqHUC+F8AVnm7TX7eKb++QHWse+Rz3gfgI8U+t6TzvACJLJ2SPzf5HN6Xf3YJnZgN38mpfqi0AkEQRJkwGy0dgiAIIgUk+ARBEGUCCT5BEESZQIJPEARRJpDgEwRBlAkk+ARBEGUCCT5BEESZ8P8BqcqOBppbv2MAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "f, Pxx_den = signal.welch(data, fs, nperseg=1024)\n",
    "plt.semilogy(f, Pxx_den)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Record Speech Examples:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 120,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "\n",
    "fs = 44100  # Sample rate\n",
    "seconds = 30  # Duration of recording\n",
    "\n",
    "myrecording = sd.rec(int(seconds * fs), samplerate=fs, channels=2)\n",
    "sd.wait()  # Wait until recording is finished"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 121,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Save as WAV file - note do this twice once for speech & once for non-speech\n",
    "wavfile.write('TrainingData/non_speech.wav', fs, myrecording)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "fs = 44100  # Sample rate\n",
    "duration = 0.2  # seconds\n",
    "sound_recording = sd.rec(int(duration * fs), samplerate=fs, channels=1)\n",
    "sound_recording = np.transpose(sound_recording)[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.00922368\n"
     ]
    }
   ],
   "source": [
    "freq, power = signal.periodogram(sound_recording, fs)\n",
    "total_energy = np.sum(power)\n",
    "in_range = lambda (f) : (f > 300) and (f < 2000)\n",
    "band_energy = np.sum([power[i] for i in range(len(power)) if in_range(freq[i])]) \n",
    "print band_energy / total_energy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 113,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.3840417e-22"
      ]
     },
     "execution_count": 113,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "power[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYYAAAD8CAYAAABzTgP2AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nrkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x112e3e710>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(sound_recording)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Train Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Load Sounds:\n",
    "fs = 44100  # Sample rate\n",
    "duration = 0.2\n",
    "[fs, speech_recording] = wavfile.read('TrainingData/speech.wav', )\n",
    "[fs, ambient_recording] = wavfile.read('TrainingData/non_speech.wav')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "10"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.size(speech_recording[1000:1010,0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "def Sound2Blocks(x, fs, block_duration):\n",
    "    sample_length = int(fs * block_duration)\n",
    "    n_samples = int(len(x) / sample_length)\n",
    "    samples = []\n",
    "    for i in range(n_samples):\n",
    "        block = x[(i*sample_length):((i+1)*sample_length),0]\n",
    "        samples.append(block)\n",
    "    return samples"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "def ExtractPowerBands(x, fs, band_bins):\n",
    "    freq, power = signal.periodogram(x, fs)\n",
    "    total_energy = np.sum(power)\n",
    "    band_energy_fraction = [0.0] * (len(band_bins))\n",
    "    band_ind = 0\n",
    "    for i in range(len(power)):\n",
    "        if(freq[i]>band_bins[band_ind]):\n",
    "            band_ind += 1\n",
    "        band_energy_fraction[band_ind] += power[i]\n",
    "    normalized_band_energy = [b / total_energy for b in band_energy_fraction]\n",
    "    return normalized_band_energy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "speech_blocks = Sound2Blocks(speech_recording, fs, duration)\n",
    "non_speech_blocks = Sound2Blocks(ambient_recording,fs, duration)\n",
    "band_bins = [180, 250, 360, 500, 1000, 2000, 5000, 10000, 25000]\n",
    "X = []\n",
    "prior_bands = [0.0]*len(band_bins)\n",
    "for block in speech_blocks:\n",
    "    bands = ExtractPowerBands(block, fs, band_bins)\n",
    "    if(prior_bands[0] != 0.0):\n",
    "        bands.extend(prior_bands)\n",
    "        X.append(bands)\n",
    "    prior_bands = bands[0:len(band_bins)]\n",
    "    \n",
    "prior_bands = [0.0]*len(band_bins)\n",
    "for block in non_speech_blocks:\n",
    "    bands = ExtractPowerBands(block, fs, band_bins)\n",
    "    if(prior_bands[0] != 0.0):\n",
    "        bands.extend(prior_bands)\n",
    "        X.append(bands)\n",
    "    prior_bands = bands[0:len(band_bins)]\n",
    "is_speech = [True] * (len(speech_blocks) - 1)\n",
    "isnot_speech = [False] * (len(non_speech_blocks) - 1)\n",
    "is_speech.extend(isnot_speech)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.9026845637583892\n"
     ]
    }
   ],
   "source": [
    "clf = SVC()\n",
    "clf.fit(X, is_speech)\n",
    "print(clf.score(X, is_speech))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "c_range = np.logspace(-2,10,13)\n",
    "gamma_range = np.logspace(-9,3,13)\n",
    "parameter_grid = dict(gamma = gamma_range, C = c_range)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "GridSearchCV(cv=StratifiedShuffleSplit(n_splits=3, random_state=None, test_size=0.25,\n",
       "            train_size=None),\n",
       "             error_score=nan,\n",
       "             estimator=SVC(C=1.0, break_ties=False, cache_size=200,\n",
       "                           class_weight=None, coef0=0.0,\n",
       "                           decision_function_shape='ovr', degree=3,\n",
       "                           gamma='scale', kernel='rbf', max_iter=-1,\n",
       "                           probability=False, random_state=None, shrinking=True,\n",
       "                           tol=0.001, verbose=False),\n",
       "             iid='deprecated', n_jobs=None,\n",
       "             param_grid={'C': array([1.e-02, 1.e-01, 1.e+00, 1.e+01, 1.e+02, 1.e+03, 1.e+04, 1.e+05,\n",
       "       1.e+06, 1.e+07, 1.e+08, 1.e+09, 1.e+10]),\n",
       "                         'gamma': array([1.e-09, 1.e-08, 1.e-07, 1.e-06, 1.e-05, 1.e-04, 1.e-03, 1.e-02,\n",
       "       1.e-01, 1.e+00, 1.e+01, 1.e+02, 1.e+03])},\n",
       "             pre_dispatch='2*n_jobs', refit=True, return_train_score=False,\n",
       "             scoring=make_scorer(precision_score), verbose=0)"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cv = StratifiedShuffleSplit(n_splits = 3, test_size= 0.25)\n",
    "scoring_metric = metrics.make_scorer(metrics. precision_score)\n",
    "grid = GridSearchCV(SVC(), param_grid=parameter_grid, cv = cv, scoring=scoring_metric)\n",
    "grid.fit(X,is_speech)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'C': 1000.0, 'gamma': 0.1}"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "grid.best_params_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8611345436762639"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "grid.best_score_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "SVC(C=1000.0, break_ties=False, cache_size=200, class_weight=None, coef0=0.0,\n",
       "    decision_function_shape='ovr', degree=3, gamma=0.1, kernel='rbf',\n",
       "    max_iter=-1, probability=False, random_state=None, shrinking=True,\n",
       "    tol=0.001, verbose=False)"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "grid.best_estimator_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "scores = grid.cv_results_['mean_test_score'].reshape(len(c_range), len(gamma_range))\n",
    "plt.imshow(scores)\n",
    "plt.xlabel('gamma')\n",
    "plt.ylabel('C')\n",
    "plt.xticks(np.arange(len(gamma_range)), gamma_range, rotation = 45)\n",
    "plt.yticks(np.arange(len(c_range)), c_range)\n",
    "plt.colorbar()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Save Model:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "False"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "band_bins = [180, 250, 360, 500, 1000, 2000, 5000, 10000, 25000]\n",
    "prior_bands = [0.0]*len(band_bins)\n",
    "clf = grid.best_estimator_\n",
    "clf.fit(X,is_speech)\n",
    "clf.predict(np.reshape([0.0]*2*len(band_bins),(1,18)))[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 256,
   "metadata": {},
   "outputs": [],
   "source": [
    "#with open('svm_classifier.pkl', 'wb') as fid:\n",
    "#    cPickle.dump(clf, fid)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "filename = 'svm_classifier_2.pkl'\n",
    "pickle.dump(clf, open(filename, 'wb'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Record Laughter"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "fs = 44100  # Sample rate\n",
    "seconds = 10  # Duration of recording\n",
    "\n",
    "laughter_recording = sd.rec(int(seconds * fs), samplerate=fs, channels=1)\n",
    "sd.wait()  # Wait until recording is finished\n",
    "wavfile.write('laughter.wav', fs, laughter_recording)  # Save as WAV file "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "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": "3.6.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}