convert-en-m-adapt1.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'\\nThis program makes learning ev-gmm.\\n'"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# -*- coding: utf-8 -*-\n",
    "\n",
    "\"\"\"\n",
    "This program makes learning ev-gmm.\n",
    "\"\"\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# __future__ module make compatible python2 and python3\n",
    "from __future__ import division, print_function\n",
    "\n",
    "# basic modules\n",
    "import os\n",
    "import os.path\n",
    "import time\n",
    "\n",
    "# for warning ignore\n",
    "import warnings\n",
    "#warning.filterwarnings('ignore')\n",
    "\n",
    "# for file system manupulation \n",
    "from shutil import rmtree \n",
    "import glob\n",
    "import argparse\n",
    "\n",
    "# for save object\n",
    "import pickle\n",
    "\n",
    "# for make glaph\n",
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "sns.set_style(\"whitegrid\")\n",
    "plt.rcParams['figure.figsize'] = (16, 5)\n",
    "import librosa.display\n",
    "\n",
    "# for scientific computing\n",
    "import numpy as np\n",
    "from numpy.linalg import norm \n",
    "from sklearn.decomposition import PCA\n",
    "from sklearn.mixture import GMM # GMM class cannot use after sklearn 0.20.0\n",
    "import sklearn.mixture\n",
    "#from sklearn.mixture.gaussian_mixture import _compute_precision_cholesky\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "import scipy.sparse\n",
    "from scipy.signal import firwin, lfilter\n",
    "\n",
    "# for display audio controler\n",
    "from IPython.display import Audio\n",
    "\n",
    "# for manuplate audio data\n",
    "import soundfile as sf\n",
    "import pyworld as pw\n",
    "import pysptk\n",
    "from dtw import dtw\n",
    "from fastdtw import fastdtw"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "class WORLD(object):\n",
    "    \"\"\"\n",
    "    WORLD based speech analyzer and synthezer.\n",
    "    \n",
    "    Ref : https://github.com/k2kobayashi/sprocket/\n",
    "    \"\"\"\n",
    "    def __init__(self, fs=16000, fftl=1024, shiftms=5.0, minf0=40.0, maxf0=500.0):\n",
    "        \"\"\"\n",
    "        Parameters\n",
    "        ----------\n",
    "        fs : int \n",
    "            Sampling frequency \n",
    "        fftl : int\n",
    "            FFT length\n",
    "        shiftms : float\n",
    "            Shift length [ms]\n",
    "        minf0 : float\n",
    "            Floor in F0 estimation\n",
    "        maxf0 : float\n",
    "            Seli in F0 estimation\n",
    "        \"\"\"\n",
    "        \n",
    "        self.fs = fs\n",
    "        self.fftl = fftl\n",
    "        self.shiftms = shiftms\n",
    "        self.minf0 = minf0\n",
    "        self.maxf0 = maxf0\n",
    "        \n",
    "    def analyze(self, x):\n",
    "        \"\"\"\n",
    "        Analyze acoustic featueres.\n",
    "        \n",
    "        Parameters\n",
    "        ----------\n",
    "        x : array, shape(`T`)\n",
    "            monoral speech signal in time domain\n",
    "        \n",
    "        Returns\n",
    "        ----------\n",
    "        f0 : array, shape(`T`)\n",
    "            F0 sequence\n",
    "        sp : array, shape(`T`, `fftl / 2 + 1`)\n",
    "            Spectral envelope sequence\n",
    "        ap : array, shape(`T`, `fftl / 2 + 1`)\n",
    "            aperiodicity sequence\n",
    "        \"\"\"\n",
    "        \n",
    "        f0, time_axis = pw.harvest(x, self.fs, f0_floor=self.minf0,\n",
    "                                   f0_ceil=self.maxf0, frame_period=self.shiftms)\n",
    "        sp = pw.cheaptrick(x, f0, time_axis, self.fs, fft_size=self.fftl)\n",
    "        ap = pw.d4c(x, f0, time_axis, self.fs, fft_size=self.fftl)\n",
    "        \n",
    "        assert sp.shape == ap.shape\n",
    "        \n",
    "        return f0, sp, ap\n",
    "    \n",
    "    def analyze_f0(self, x):\n",
    "        \"\"\"\n",
    "        Analyze f0.\n",
    "        \n",
    "        Parameters\n",
    "        ----------\n",
    "        x : array, shape(`T`)\n",
    "            monoral speech signal in time domain\n",
    "        \n",
    "        Returns\n",
    "        ----------\n",
    "        f0 : array, shape(`T`)\n",
    "            F0 sequence\n",
    "        \"\"\"\n",
    "\n",
    "        f0, time_axis = pw.harvest(x, self.fs, f0_floor=self.minf0,\n",
    "                                   f0_ceil=self.maxf0, frame_period=self.shiftms)\n",
    "        \n",
    "        assert f0.shape == x.shape()\n",
    "        \n",
    "        return f0\n",
    "    \n",
    "    def synthesis(self, f0, sp, ap):\n",
    "        \"\"\"\n",
    "        Re-synthesizes a speech waveform from acoustic featueres.\n",
    "        \n",
    "        Parameters\n",
    "        ----------\n",
    "        f0 : array, shape(`T`)\n",
    "            F0 sequence\n",
    "        sp : array, shape(`T`, `fftl / 2 + 1`)\n",
    "            Spectral envelope sequence\n",
    "        ap : array, shape(`T`, `fftl / 2 + 1`)\n",
    "            aperiodicity sequence\n",
    "        \"\"\"\n",
    "\n",
    "        return pw.synthesize(f0, sp, ap, self.fs, frame_period=self.shiftms)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "class FeatureExtractor(object):\n",
    "    \"\"\"\n",
    "    Analyze acoustic features from a waveform.\n",
    "    \n",
    "    This class may have several types of estimeter like WORLD or STRAIGHT.\n",
    "    Default type is WORLD.\n",
    "    \n",
    "    Ref : https://github.com/k2kobayashi/sprocket/\n",
    "    \"\"\"\n",
    "    \n",
    "    def __init__(self, analyzer='world', fs=16000, fftl=1024, \n",
    "                 shiftms=5.0, minf0=50.0, maxf0=500.0):\n",
    "        \"\"\"\n",
    "        Parameters\n",
    "        ----------\n",
    "        analyzer : str\n",
    "            Analyzer\n",
    "        fs : int \n",
    "            Sampling frequency \n",
    "        fftl : int\n",
    "            FFT length\n",
    "        shiftms : float\n",
    "            Shift length [ms]\n",
    "        minf0 : float\n",
    "            Floor in F0 estimation\n",
    "        maxf0 : float\n",
    "            Seli in F0 estimation\n",
    "        \"\"\"\n",
    "        \n",
    "        self.analyzer = analyzer\n",
    "        self.fs = fs\n",
    "        self.fftl = fftl\n",
    "        self.shiftms = shiftms\n",
    "        self.minf0 = minf0\n",
    "        self.maxf0 = maxf0\n",
    "    \n",
    "        if self.analyzer == 'world':\n",
    "            self.analyzer = WORLD(fs=self.fs, fftl=self.fftl, \n",
    "                                  minf0=self.minf0, maxf0=self.maxf0, shiftms=self.shiftms)\n",
    "        else:\n",
    "            raise('Analyzer Error : not support type, see FeatureExtractor class.')\n",
    "        \n",
    "        self._f0 = None\n",
    "        self._sp = None\n",
    "        self._ap = None\n",
    "        \n",
    "    def analyze(self, x):\n",
    "        \"\"\"\n",
    "        Analyze acoustic featueres.\n",
    "        \n",
    "        Parameters\n",
    "        ----------\n",
    "        x : array, shape(`T`)\n",
    "            monoral speech signal in time domain\n",
    "        \n",
    "        Returns\n",
    "        ----------\n",
    "        f0 : array, shape(`T`)\n",
    "            F0 sequence\n",
    "        sp : array, shape(`T`, `fftl / 2 + 1`)\n",
    "            Spectral envelope sequence\n",
    "        ap : array, shape(`T`, `fftl / 2 + 1`)\n",
    "            aperiodicity sequence\n",
    "        \"\"\"\n",
    "        \n",
    "        self.x = np.array(x, dtype=np.float)\n",
    "        self._f0, self._sp, self._ap = self.analyzer.analyze(self.x)\n",
    "        \n",
    "        # check f0 < 0\n",
    "        self._f0[self._f0 < 0] = 0\n",
    "        \n",
    "        if np.sum(self._f0) == 0.0:\n",
    "            print(\"Warning : F0 values are all zero.\")\n",
    "        \n",
    "        return self._f0, self._sp, self._ap\n",
    "    \n",
    "    def analyze_f0(self, x):\n",
    "        \"\"\"\n",
    "        Analyze f0.\n",
    "        \n",
    "        Parameters\n",
    "        ----------\n",
    "        x : array, shape(`T`)\n",
    "            monoral speech signal in time domain\n",
    "        \n",
    "        Returns\n",
    "        ----------\n",
    "        f0 : array, shape(`T`)\n",
    "            F0 sequence\n",
    "        \"\"\"\n",
    "\n",
    "        self.x = np.array(x, dtype=np.float)\n",
    "        self._f0 = self.analyzer.analyze_f0(self.x)\n",
    "\n",
    "        # check f0 < 0\n",
    "        self._f0[self._f0 < 0] = 0\n",
    "        \n",
    "        if np.sum(self._f0) == 0.0:\n",
    "            print(\"Warning : F0 values are all zero.\")\n",
    "        \n",
    "        return self._f0\n",
    "    \n",
    "    def mcep(self, dim=24, alpha=0.42):\n",
    "        \"\"\"\n",
    "        Convert mel-cepstrum sequence from spectral envelope.\n",
    "        \n",
    "        Parameters\n",
    "        ----------\n",
    "        dim : int\n",
    "            mel-cepstrum dimension\n",
    "        alpha : float\n",
    "            parameter of all-path filter\n",
    "        \n",
    "        Returns\n",
    "        ----------\n",
    "        mcep : array, shape(`T`, `dim + 1`)\n",
    "            mel-cepstrum sequence\n",
    "        \"\"\"        \n",
    "        \n",
    "        self._analyzed_check()\n",
    "        \n",
    "        return pysptk.sp2mc(self._sp, dim, alpha)\n",
    "    \n",
    "    def codeap(self):\n",
    "        \"\"\"\n",
    "        \"\"\"\n",
    "        self._analyzed_check()\n",
    "        \n",
    "        return pw.code_aperiodicity(self._ap, self.fs)\n",
    "    \n",
    "    def npow(self):\n",
    "        \"\"\"\n",
    "        Normalized power sequence from spectral envelope.\n",
    "        \n",
    "        Returns\n",
    "        ----------\n",
    "        npow : vector, shape(`T`, `1`)\n",
    "            Normalized power sequence of the given waveform\n",
    "        \"\"\"\n",
    "        \n",
    "        self._analyzed_check()\n",
    "        \n",
    "        npow = np.apply_along_axis(self._spvec2pow, 1, self._sp)\n",
    "        \n",
    "        meanpow = np.mean(npow)\n",
    "        npow = 10.0 * np.log10(npow / meanpow)\n",
    "        \n",
    "        return npow\n",
    "    \n",
    "    def _spvec2pow(self, specvec):\n",
    "        \"\"\"\n",
    "        \"\"\"\n",
    "        fftl2 = len(specvec) - 1\n",
    "        fftl = fftl2 * 2\n",
    "        \n",
    "        power = specvec[0] + specvec[fftl2]\n",
    "        for k in range(1, fftl2):\n",
    "            power += 2.0 * specvec[k]\n",
    "        power /= fftl\n",
    "        \n",
    "        return power\n",
    "        \n",
    "    def _analyzed_check(self):\n",
    "        if self._f0 is None and self._sp is None and self._ap is None:\n",
    "            raise('Call FeatureExtractor.analyze() before this method.')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Synthesizer(object):\n",
    "    \"\"\"\n",
    "    Synthesize a waveform from acoustic features.\n",
    "    \n",
    "    Ref : https://github.com/k2kobayashi/sprocket/\n",
    "    \"\"\"\n",
    "    def __init__(self, fs=16000, fftl=1024, shiftms=5.0):\n",
    "        \"\"\"\n",
    "        Parameters\n",
    "        ----------\n",
    "        fs : int \n",
    "            Sampling frequency \n",
    "        fftl : int\n",
    "            FFT length\n",
    "        shiftms : float\n",
    "            Shift length [ms]\n",
    "        \"\"\"\n",
    "        \n",
    "        self.fs = fs\n",
    "        self.fftl = fftl\n",
    "        self.shiftms = shiftms\n",
    "    \n",
    "    def synthesis(self, f0, mcep, ap, rmcep=None, alpha=0.42):\n",
    "        \"\"\"\n",
    "        Re-synthesizes a speech waveform from acoustic featueres.\n",
    "        \n",
    "        Parameters\n",
    "        ----------\n",
    "        f0 : array, shape(`T`)\n",
    "            F0 sequence\n",
    "        mcep : array, shape(`T`, `dim`)\n",
    "            mel-cepstrum sequence\n",
    "        ap : array, shape(`T`, `fftl / 2 + 1`)\n",
    "            aperiodicity sequence\n",
    "        rmcep : array, shape(`T`, `dim`)\n",
    "            array of reference mel-cepstrum sequence\n",
    "        alpha : float\n",
    "            parameter of all-path filter\n",
    "            \n",
    "        Returns\n",
    "        ----------\n",
    "        wav : array,\n",
    "            syntesized waveform\n",
    "        \"\"\"\n",
    "        \n",
    "        if rmcep is not None:\n",
    "            # power modification\n",
    "            mcep = mod_power(mcep, rmcep, alpha=alpha)\n",
    "        \n",
    "        sp = pysptk.mc2sp(mcep, alpha, self.fftl)\n",
    "        wav = pw.synthesize(f0, sp, ap, self.fs, frame_period=self.shiftms)\n",
    "        \n",
    "        return wav\n",
    "        \n",
    "    def synthesis_diff(self, x, diffmcep, rmcep=None, alpha=0.42):\n",
    "        \"\"\"\n",
    "        Re-synthesizes a speech waveform from acoustic featueres.\n",
    "        filtering with a differential mel-cepstrum.\n",
    "        \n",
    "        Parameters\n",
    "        ----------\n",
    "        x : array, shape(`samples`)\n",
    "            array of waveform sequence\n",
    "        diffmcep : array, shape(`T`, `dim`)\n",
    "            array of differential mel-cepstrum sequence\n",
    "        rmcep : array, shape(`T`, `dim`)\n",
    "            array of reference mel-cepstrum sequence\n",
    "        alpha : float\n",
    "            parameter of all-path filter\n",
    "            \n",
    "        Returns\n",
    "        ----------\n",
    "        wav : array,\n",
    "            syntesized waveform\n",
    "        \"\"\"      \n",
    "        \n",
    "        x = x.astype(np.float64)\n",
    "        dim = diffmcep.shape[1] - 1 \n",
    "        shiftl = int(self.fs / 1000 * self.shiftms)\n",
    "        \n",
    "        if rmcep is not None:\n",
    "            # power modification\n",
    "            diffmcep = mod_power(rmcep + diffmcep, rmcep, alpha=alpha) - rmcep        \n",
    "        \n",
    "        # mc2b = transform mel-cepstrum to MLSA digital filter coefficients.\n",
    "        b = np.apply_along_axis(pysptk.mc2b, 1, diffmcep, alpha)\n",
    "        \n",
    "        mlsa_fil = pysptk.synthesis.Synthesizer(pysptk.synthesis.MLSADF(dim, alpha=alpha),\n",
    "                                                shiftl)\n",
    "        wav = mlsa_fil.synthesis(x, b)\n",
    "        \n",
    "        return wav\n",
    "    \n",
    "    def synthesis_sp(self, f0, sp, ap):\n",
    "        \"\"\"\n",
    "        Re-synthesizes a speech waveform from acoustic featueres.\n",
    "        \n",
    "        Parameters\n",
    "        ----------\n",
    "        f0 : array, shape(`T`)\n",
    "            F0 sequence\n",
    "        spc : array, shape(`T`, `dim`)\n",
    "            mel-cepstrum sequence\n",
    "        ap : array, shape(`T`, `fftl / 2 + 1`)\n",
    "            aperiodicity sequence\n",
    "            \n",
    "        Returns\n",
    "        ----------\n",
    "        wav : array,\n",
    "            syntesized waveform\n",
    "        \"\"\"      \n",
    "        \n",
    "        wav = pw.synthesize(f0, sp, ap, self.fs, frame_period=self.shiftms)\n",
    "        \n",
    "        return wav\n",
    "    \n",
    "def mod_power(cvmcep, rmcep, alpha=0.42, irlen=256):\n",
    "    \"\"\"\n",
    "    power modification based on inpuulse responce\n",
    "\n",
    "    Parameters\n",
    "    ----------\n",
    "    cvmcep : array, shape(`T`, `dim`)\n",
    "        array of converted mel-cepstrum\n",
    "    rmcep : arraym shape(`T`, `dim`)\n",
    "        array of reference mel-cepstrum\n",
    "    alpha : float\n",
    "        parameter of all-path filter\n",
    "    irlen : int\n",
    "        Length for IIR filter\n",
    "\n",
    "    Returns\n",
    "    ----------\n",
    "    modified_cvmcep : array, shape(`T`, `dim`)\n",
    "        array of power modified converted mel-cepstrum\n",
    "    \"\"\"\n",
    "\n",
    "    if rmcep.shape != cvmcep.shape:\n",
    "        raise ValueError(\n",
    "            \"The shape of the converted and reference mel-cepstrum are different : {} / {}.format(cvmcep.shape, rmcep.shape)\"\n",
    "        )\n",
    "\n",
    "    # mc2e = Compute energy from mel-cepstrum. e-option\n",
    "    cv_e = pysptk.mc2e(cvmcep, alpha=alpha, irlen=irlen)\n",
    "    r_e = pysptk.mc2e(rmcep, alpha=alpha, irlen=irlen)\n",
    "\n",
    "    dpow = np.log(r_e / cv_e) / 2\n",
    "\n",
    "    modified_cvmcep = np.copy(cvmcep)\n",
    "    modified_cvmcep[:, 0] += dpow\n",
    "\n",
    "    return modified_cvmcep"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# def util methods\n",
    "def melcd(array1, array2):\n",
    "    \"\"\"\n",
    "    calculate mel-cepstrum distortion\n",
    "    \n",
    "    Parameters\n",
    "    ----------\n",
    "    array1, array2 : array, shape(`T`, `dim`) or shape(`dim`)\n",
    "        Array of original and target.\n",
    "    \n",
    "    Returns\n",
    "    ----------\n",
    "    mcd : scala, number > 0\n",
    "        Scala of mel-cepstrum distoriton\n",
    "    \"\"\"\n",
    "    if array1.shape != array2.shape:\n",
    "        raise ValueError(\n",
    "            \"The shape of both array are different : {} / {}.format(array1.shape,array2.shape)\"\n",
    "        )    \n",
    "   \n",
    "    if array1.ndim == 2:\n",
    "        diff = array1 - array2\n",
    "        mcd = 10.0 / np.log(10) * np.mean(np.sqrt(2.0 * np.sum(diff ** 2, axis=1)))\n",
    "    elif array1.ndim == 1:\n",
    "        diff = array1 - array2\n",
    "        mcd = 10.0 / np.log(10) * np.sqrt(2.0 * np.sum(diff ** 2))\n",
    "    else:\n",
    "        raise ValueError(\"Dimension mismatch.\")\n",
    "        \n",
    "    return mcd\n",
    "\n",
    "def delta(data, win=[-1.0, 1.0, 0]):\n",
    "    \"\"\"\n",
    "    calculate delta component\n",
    "    \n",
    "    Parameters\n",
    "    ----------\n",
    "    data : array, shape(`T`, `dim`)\n",
    "        Array of static matrix sequence.\n",
    "    win : array, shape(`3`)\n",
    "        The shape of window matrix.\n",
    "    \n",
    "    Returns\n",
    "    ----------\n",
    "    delta : array, shape(`T`, `dim`)\n",
    "        Array of delta matrix sequence.\n",
    "    \"\"\"\n",
    "    \n",
    "    if data.ndim == 1:\n",
    "        # change vector into 1d-array\n",
    "        T = len(data)\n",
    "        dim = data.ndim\n",
    "        data = data.reshape(T, dim)\n",
    "    else:\n",
    "        T, dim = data.shape\n",
    "    \n",
    "    win = np.array(win, dtype=np.float64)\n",
    "    delta = np.zeros((T, dim))\n",
    "    \n",
    "    delta[0] = win[0] * data[0] + win[1] * data[1]\n",
    "    delta[-1] = win[0] * data[-2] + win[1] * data[-1]\n",
    "    \n",
    "    for i in range(len(win)):\n",
    "        delta[1:T - 1] += win[i] * delta[i:T - 2 + i]\n",
    "    \n",
    "    return delta\n",
    "\n",
    "def static_delta(data, win=[-1.0, 1.0, 0]):\n",
    "    \"\"\"\n",
    "    calculate static and delta component\n",
    "    \n",
    "    Parameters\n",
    "    ----------\n",
    "    data : array, shape(`T`, `dim`)\n",
    "        Array of static matrix sequence.\n",
    "    win : array, shape(`3`)\n",
    "        The shape of window matrix.\n",
    "    \n",
    "    Returns\n",
    "    ----------\n",
    "    sddata : array, shape(`T`, `dim * 2`)\n",
    "        Array of static and delta matrix sequence.\n",
    "    \"\"\"\n",
    "    \n",
    "    sddata = np.c_[data, delta(data, win)]\n",
    "    \n",
    "    assert sddata.shape[1] == data.shape[1] * 2\n",
    "    \n",
    "    return sddata\n",
    "\n",
    "def construct_static_and_delta_matrix(T, D, win=[-1.0, 1.0, 0]):\n",
    "    \"\"\"\n",
    "    calculate static and delta transformation matrix\n",
    "    \n",
    "    Parameters\n",
    "    ----------\n",
    "    T : scala, `T`\n",
    "        Scala of time length\n",
    "    D : scala, `D`\n",
    "        Scala of the number of dimension.\n",
    "    win : array, shape(`3`)\n",
    "        The shape of window matrix.\n",
    "    \n",
    "    Returns\n",
    "    ----------\n",
    "    W : array, shape(`2 * D * T`, `D * T`)\n",
    "        Array of static and delta transformation matrix.\n",
    "    \"\"\"\n",
    "    \n",
    "    static = [0, 1, 0]\n",
    "    delta = win\n",
    "    assert len(static) == len(delta)\n",
    "    \n",
    "    # generate full W\n",
    "    DT = D * T\n",
    "    ones = np.ones(DT)\n",
    "    row = np.arange(2 * DT).reshape(2 * T, D) # generate serial numbers\n",
    "    static_row = row[::2] # [1,2,3,4,5] => [1,3,5]\n",
    "    delta_row = row[1::2] # [1,2,3,4,5] => [2,4]\n",
    "    col = np.arange(DT)\n",
    "    \n",
    "    data = np.array([ones * static[0], ones * static[1],\n",
    "                      ones * static[2], ones * delta[0],\n",
    "                      ones * delta[1], ones * delta[2]]).flatten()\n",
    "    row = np.array([[static_row] * 3, [delta_row] * 3]).flatten()\n",
    "    col = np.array([[col - D, col, col + D] * 2]).flatten()\n",
    "\n",
    "    # remove component at first and end frame\n",
    "    valid_idx = np.logical_not(np.logical_or(col < 0, col >= DT))\n",
    "    \n",
    "    W = scipy.sparse.csr_matrix(\n",
    "        (data[valid_idx], (row[valid_idx], col[valid_idx])), shape=(2 * DT, DT))\n",
    "    W.eliminate_zeros()\n",
    "    \n",
    "    return W\n",
    "    \n",
    "def extfrm(data, npow, power_threshold=-20):\n",
    "    \"\"\"\n",
    "    Extract frame over the power threshold\n",
    "    \n",
    "    Parameters\n",
    "    ----------\n",
    "    data : array, shape(`T`, `dim`)\n",
    "        array of input data\n",
    "    npow : array, shape(`T`)\n",
    "        vector of normalized power sequence\n",
    "    threshold : scala\n",
    "        scala of power threshold [dB]\n",
    "        \n",
    "    Returns\n",
    "    ----------\n",
    "    data : array, shape(`T_ext`, `dim`)\n",
    "        remaining data after extracting frame\n",
    "        `T_ext` <= `T`\n",
    "    \"\"\"\n",
    "    T = data.shape[0]\n",
    "    if T != len(npow):\n",
    "        raise(\"Length of two vectors is different.\")\n",
    "        \n",
    "    valid_index = np.where(npow > power_threshold)\n",
    "    extdata = data[valid_index]\n",
    "    assert extdata.shape[0] <= T\n",
    "    \n",
    "    return extdata\n",
    "\n",
    "def estimate_twf(orgdata, tardata, distance='melcd', fast=True, otflag=None):\n",
    "    \"\"\"\n",
    "    time warping function estimator\n",
    "    \n",
    "    Parameters\n",
    "    ----------\n",
    "    orgdata : array, shape(`T_org`, `dim`)\n",
    "        array of source feature\n",
    "    tardata : array, shape(`T_tar`, `dim`)\n",
    "        array of target feature\n",
    "    distance : str \n",
    "        distance function\n",
    "    fast : bool\n",
    "        use fastdtw instead of dtw\n",
    "    otflag : str\n",
    "        Alignment into the length of specification\n",
    "        'org' : alignment into original length\n",
    "        'tar' : alignment into target length   \n",
    "    \n",
    "    Returns\n",
    "    ----------\n",
    "    twf : array, shape(`2`, `T`)\n",
    "        time warping function between original and target\n",
    "    \"\"\"\n",
    "    \n",
    "    if distance == 'melcd':\n",
    "        def distance_func(x, y): return melcd(x, y)\n",
    "    else:\n",
    "        raise ValueError('this distance method is not support.')\n",
    "    \n",
    "    if fast:\n",
    "        _, path = fastdtw(orgdata, tardata, dist=distance_func)\n",
    "        twf = np.array(path).T\n",
    "    else:\n",
    "        _, _, _, twf = dtw(orgdata, tardata, distance_func) \n",
    "        \n",
    "    if otflag is not None:\n",
    "        twf = modify_twf(twf, otflag=otflag)\n",
    "    \n",
    "    return twf\n",
    "\n",
    "def align_data(org_data, tar_data, twf):\n",
    "    \"\"\"\n",
    "    get aligned joint feature vector\n",
    "    \n",
    "    Parameters\n",
    "    ----------\n",
    "    org_data : array, shape(`T_org`, `dim_org`)\n",
    "        Acoustic feature vector of original speaker\n",
    "    tar_data : array, shape(`T_tar`, `dim_tar`)\n",
    "        Acoustic feature vector of target speaker\n",
    "    twf : array, shape(`2`, `T`)\n",
    "        time warping function between original and target\n",
    "        \n",
    "    Returns\n",
    "    ----------\n",
    "    jdata : array, shape(`T_new`, `dim_org + dim_tar`)\n",
    "        Joint feature vector between source and target\n",
    "    \"\"\"\n",
    "    \n",
    "    jdata = np.c_[org_data[twf[0]], tar_data[twf[1]]]\n",
    "    return jdata\n",
    "\n",
    "def modify_twf(twf, otflag=None):\n",
    "    \"\"\"\n",
    "    align specified length\n",
    "    \n",
    "    Parameters\n",
    "    ----------\n",
    "    twf : array, shape(`2`, `T`)\n",
    "        time warping function between original and target\n",
    "    otflag : str\n",
    "        Alignment into the length of specification\n",
    "        'org' : alignment into original length\n",
    "        'tar' : alignment into target length   \n",
    "    \n",
    "    Returns\n",
    "    ----------\n",
    "    mod_twf : array, shape(`2`, `T_new`)\n",
    "        time warping function of modified alignment\n",
    "    \"\"\"\n",
    "    \n",
    "    if otflag == 'org':\n",
    "        of, indice = np.unique(twf[0], return_index=True)\n",
    "        mod_twf = np.c_[of, twf[1][indice]].T\n",
    "    elif otflag == 'tar':\n",
    "        tf, indice = np.unique(twf[1], return_index=True)\n",
    "        mod_twf = np.c_[twf[0][indice], tf].T \n",
    "    \n",
    "    return mod_twf\n",
    "\n",
    "def low_cut_filter(x, fs, cutoff=70):\n",
    "    \"\"\"\n",
    "    low cut filter\n",
    "    \n",
    "    Parameters\n",
    "    ----------\n",
    "    x : array, shape('samples')\n",
    "        waveform sequence\n",
    "    fs : array, int\n",
    "        Sampling frequency\n",
    "    cutoff : float\n",
    "        cutoff frequency of low cut filter\n",
    "    \n",
    "    Returns\n",
    "    ----------\n",
    "    lct_x : array, shape('samples')\n",
    "        Low cut filtered waveform sequence\n",
    "    \"\"\"\n",
    "    \n",
    "    nyquist = fs // 2\n",
    "    norm_cutoff = cutoff / nyquist\n",
    "    \n",
    "    # low cut filter\n",
    "    fil = firwin(255, norm_cutoff, pass_zero=False)\n",
    "    lct_x = lfilter(fil, 1, x)\n",
    "    \n",
    "    return lct_x\n",
    "\n",
    "def extsddata(data, npow, power_threshold=-20):\n",
    "    \"\"\"\n",
    "    get power extract static and delta feature vector\n",
    "    \n",
    "    Parameters\n",
    "    ----------\n",
    "    data : array, shape(`T`, `dim`)\n",
    "        acoustic feature vector\n",
    "    npow : array, shape(`T`)\n",
    "        normalized power vector\n",
    "    power_threshold : float\n",
    "        power threshold\n",
    "    \n",
    "    Returns\n",
    "    ----------\n",
    "    extsddata : array, shape(`T_new`, `dim * 2`)\n",
    "        silence remove static and delta feature vector\n",
    "    \"\"\"\n",
    "    \n",
    "    extsddata = extfrm(static_delta(data), npow, power_threshold=power_threshold)\n",
    "    return extsddata\n",
    "\n",
    "def transform_jnt(array_list):\n",
    "    num_files = len(array_list)\n",
    "    for i in range(num_files):\n",
    "        if i == 0:\n",
    "            jnt = array_list[i]\n",
    "        else:\n",
    "            jnt = np.r_[jnt, array_list[i]]\n",
    "    return jnt\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "class F0statistics(object):\n",
    "    \"\"\"\n",
    "    Estimate F0 statistics and convert F0\n",
    "    \"\"\"\n",
    "    def __init__(self):\n",
    "        pass\n",
    "    \n",
    "    def estimate(self, f0list):\n",
    "        \"\"\"\n",
    "        estimate F0 statistics from list of f0\n",
    "        \n",
    "        Parameters\n",
    "        ----------\n",
    "        f0list : list, shape(`f0num`)\n",
    "            List of several F0 sequence\n",
    "        \n",
    "        Returns\n",
    "        ----------\n",
    "        f0stats : array, shape(`[mean, std]`)\n",
    "            values of mean and standard deviation for log f0\n",
    "        \"\"\"\n",
    "        \n",
    "        n_files = len(f0list)\n",
    "        assert n_files != 0\n",
    "        for i in range(n_files):\n",
    "            f0 = f0list[i]\n",
    "            nonzero_indices = np.nonzero(f0)\n",
    "            if i == 0:\n",
    "                f0s = np.log(f0[nonzero_indices])\n",
    "            else:\n",
    "                f0s = np.r_[f0s, np.log(f0[nonzero_indices])]\n",
    "        \n",
    "        f0stats = np.array([np.mean(f0s), np.std(f0s)])\n",
    "        \n",
    "        return f0stats\n",
    "\n",
    "    def convert(self, f0, orgf0stats, tarf0stats):\n",
    "        \"\"\"\n",
    "        convert F0 based on F0 statistics\n",
    "        \n",
    "        Parameters\n",
    "        ----------\n",
    "        f0 : array, shape(`T`, `1`)\n",
    "            array of F0 sequence\n",
    "        orgf0stats : array, shape(`[mean, std]`)\n",
    "            vectors of mean and standard deviation of log f0 for original speaker\n",
    "        tarf0stats : array, shape(`[mean, std]`)\n",
    "            vectors of mean and standard deviation of log f0 for target speaker\n",
    "        \n",
    "        Returns\n",
    "        ----------\n",
    "        cvf0 : array, shape(`T`, `1`)\n",
    "            array of converted F0 sequence\n",
    "        \"\"\"\n",
    "        \n",
    "        # get length and dimension\n",
    "        T = len(f0)\n",
    "        \n",
    "        # perform f0 conversion\n",
    "        cvf0 = np.zeros(T)\n",
    "        \n",
    "        nonzero_indices  = f0 > 0\n",
    "        cvf0[nonzero_indices] = np.exp((tarf0stats[1] / orgf0stats[1]) * (np.log(f0[nonzero_indices]) - orgf0stats[0]) + tarf0stats[0])\n",
    "        \n",
    "        return cvf0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "class GV(object):\n",
    "    \"\"\"\n",
    "    Estimate statistics and perform postfilter based on the GV statistics.\n",
    "    \"\"\"\n",
    "    def __init__(self):\n",
    "        pass\n",
    "    \n",
    "    def estimate(self, datalist):\n",
    "        \"\"\"\n",
    "        estimate GV statistics from list of data\n",
    "        \n",
    "        Parameters\n",
    "        ----------\n",
    "        datalist : list, shape(`num_data`)\n",
    "            List of several data ([T, dim]) sequence\n",
    "        \n",
    "        Returns\n",
    "        ----------\n",
    "        gvstats : array, shape(`2`, `dim`)\n",
    "            array of mean and standard deviation for GV\n",
    "        \"\"\"\n",
    "        \n",
    "        n_files = len(datalist)\n",
    "        assert n_files != 0\n",
    "        var = []\n",
    "        for i in range(n_files):\n",
    "            data = datalist[i]\n",
    "            var.append(np.var(data, axis=0))\n",
    "            \n",
    "        # calculate vm and vv\n",
    "        vm = np.mean(np.array(var), axis=0)\n",
    "        vv = np.var(np.array(var), axis=0)\n",
    "        gvstats = np.r_[vm, vv]\n",
    "        gvstats = gvstats.reshape(2, len(vm))\n",
    "        \n",
    "        return gvstats\n",
    "\n",
    "    def postfilter(self, data, gvstats, cvgvstats=None, alpha=1.0, startdim=1):\n",
    "        \"\"\"\n",
    "        perform postfilter based on GV statistics into data\n",
    "        \n",
    "        Parameters\n",
    "        ----------\n",
    "        data : array, shape(`T`, `dim`)\n",
    "            array of data sequence\n",
    "        gvstats : array, shape(`2`, `dim`)\n",
    "            array of mean and variance for target GV\n",
    "        cvgvstats : array, shape(`2`, `dim`)\n",
    "            array of mean and variance for converted GV\n",
    "        alpha : float\n",
    "            morphing coefficient between GV transformed data and data.\n",
    "            alpha * gvpf(data) + (1 - alpha) * data\n",
    "        startdim : int\n",
    "            start dimension to perform GV postfilter\n",
    "        \n",
    "        Returns\n",
    "        ----------\n",
    "        filtered_data : array, shape(`T`, `data`)\n",
    "            array of GV postfiltered data sequnece\n",
    "        \"\"\"\n",
    "        \n",
    "        # get length and dimension\n",
    "        T, dim = data.shape\n",
    "        assert gvstats is not None\n",
    "        assert dim == gvstats.shape[1]\n",
    "        \n",
    "        # calculate statics of input data\n",
    "        datamean = np.mean(data, axis=0)\n",
    "        \n",
    "        if cvgvstats is None:\n",
    "            # use variance of the given data\n",
    "            datavar = np.var(data, axis=0)\n",
    "        else:\n",
    "            # use variance of trained gv stats\n",
    "            datavar = cvgvstats[0]\n",
    "        \n",
    "        # perform GV postfilter\n",
    "        filterd = np.sqrt(gvstats[0, startdim:] / datavar[startdim:]) * (data[:, startdim:] - datamean[startdim:]) + datamean[startdim:]\n",
    "        \n",
    "        filterd_data = np.c_[data[:, :startdim], filterd]\n",
    "        \n",
    "        return alpha * filterd_data + (1 - alpha) * data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 0. config path\n",
    "__versions = \"pre-stored-en\"\n",
    "__same_path = \"./utterance/\" + __versions + \"/\"\n",
    "prepare_path = __same_path + \"output/\"\n",
    "pre_stored_source_list = __same_path + 'pre-source/**/V01/T01/**/*.wav'\n",
    "pre_stored_list = __same_path + \"pre/**/V01/T01/**/*.wav\"\n",
    "output_path = \"./utterance/old-test/orl/en-m/4/adapt1/\"\n",
    "\n",
    "# 1. estimate features\n",
    "feat = FeatureExtractor()\n",
    "synthesizer = Synthesizer()\n",
    "\n",
    "org_f0list = None\n",
    "org_splist = None\n",
    "org_mceplist = None\n",
    "org_aplist = None\n",
    "org_npowlist = None\n",
    "org_codeaplist = None\n",
    "\n",
    "if os.path.exists(prepare_path + \"_org_f0.pickle\") \\\n",
    "    and os.path.exists(prepare_path + \"_org_sp.pickle\") \\\n",
    "    and os.path.exists(prepare_path + \"_org_ap.pickle\") \\\n",
    "    and os.path.exists(prepare_path + \"_org_mcep.pickle\") \\\n",
    "    and os.path.exists(prepare_path + \"_org_npow.pickle\") \\\n",
    "    and os.path.exists(prepare_path + \"_org_codeap.pickle\"):\n",
    "        \n",
    "    with open(prepare_path + \"_org_f0.pickle\", 'rb') as f:    \n",
    "        org_f0list = pickle.load(f)\n",
    "    with open(prepare_path + \"_org_sp.pickle\", 'rb') as f:   \n",
    "        org_splist = pickle.load(f)\n",
    "    with open(prepare_path + \"_org_ap.pickle\", 'rb') as f:  \n",
    "        org_aplist = pickle.load(f)\n",
    "    with open(prepare_path + \"_org_mcep.pickle\", 'rb') as f:  \n",
    "        org_mceplist = pickle.load(f)\n",
    "    with open(prepare_path + \"_org_npow.pickle\", 'rb') as f:  \n",
    "        org_npowlist = pickle.load(f)\n",
    "    with open(prepare_path + \"_org_codeap.pickle\", 'rb') as f:  \n",
    "        org_codeaplist = pickle.load(f) \n",
    "else:\n",
    "    org_f0list = []\n",
    "    org_splist = []\n",
    "    org_mceplist = []\n",
    "    org_aplist = []\n",
    "    org_npowlist = []\n",
    "    org_codeaplist = []\n",
    "    ite = 0\n",
    "    for files in sorted(glob.iglob(pre_stored_source_list, recursive=True)):\n",
    "        wavf = files\n",
    "        x, fs = sf.read(wavf)\n",
    "        x = np.array(x, dtype=np.float)\n",
    "        x = low_cut_filter(x, fs, cutoff=70)\n",
    "        assert fs == 16000\n",
    "\n",
    "        print(\"extract acoustic featuers: \" + wavf)\n",
    "\n",
    "        f0, sp, ap = feat.analyze(x)\n",
    "        mcep = feat.mcep()\n",
    "        npow = feat.npow()\n",
    "        codeap = feat.codeap()\n",
    "        wav = synthesizer.synthesis_sp(f0, sp, ap)\n",
    "        wav = np.clip(wav, -32768, 32767)\n",
    "        sf.write(prepare_path + \"src_ansys_{}_.wav\".format(ite), wav, fs)\n",
    "\n",
    "        org_f0list.append(f0)\n",
    "        org_splist.append(sp)\n",
    "        org_mceplist.append(mcep)\n",
    "        org_aplist.append(ap)\n",
    "        org_npowlist.append(npow)\n",
    "        org_codeaplist.append(codeap)\n",
    "\n",
    "        wav = synthesizer.synthesis(f0, mcep, ap)\n",
    "        wav = np.clip(wav, -32768, 32767)\n",
    "        sf.write(prepare_path + \"src_mcep_{}_.wav\".format(ite), wav, fs)\n",
    "        ite = ite + 1\n",
    "\n",
    "    with open(prepare_path + \"_org_f0.pickle\", 'wb') as f:    \n",
    "        pickle.dump(org_f0list, f)\n",
    "    with open(prepare_path + \"_org_sp.pickle\", 'wb') as f:   \n",
    "        pickle.dump(org_splist, f)\n",
    "    with open(prepare_path + \"_org_npow.pickle\", 'wb') as f:   \n",
    "        pickle.dump(org_npowlist, f)\n",
    "    with open(prepare_path + \"_org_ap.pickle\", 'wb') as f:  \n",
    "        pickle.dump(org_aplist, f)\n",
    "    with open(prepare_path + \"_org_mcep.pickle\", 'wb') as f:  \n",
    "        pickle.dump(org_mceplist, f)\n",
    "    with open(prepare_path + \"_org_codeap.pickle\", 'wb') as f:  \n",
    "        pickle.dump(org_codeaplist, f) \n",
    "\n",
    "mid_f0list = None\n",
    "mid_mceplist = None\n",
    "mid_aplist = None\n",
    "mid_npowlist = None\n",
    "mid_splist = None\n",
    "mid_codeaplist = None        \n",
    "\n",
    "if os.path.exists(prepare_path + \"_mid_f0.pickle\") \\\n",
    "    and os.path.exists(prepare_path + \"_mid_sp_0_.pickle\") \\\n",
    "    and os.path.exists(prepare_path + \"_mid_ap_0_.pickle\") \\\n",
    "    and os.path.exists(prepare_path + \"_mid_mcep.pickle\") \\\n",
    "    and os.path.exists(prepare_path + \"_mid_npow.pickle\") \\\n",
    "    and os.path.exists(prepare_path + \"_mid_codeap.pickle\"):\n",
    "        \n",
    "    with open(prepare_path + \"_mid_f0.pickle\", 'rb') as f:    \n",
    "        mid_f0list = pickle.load(f)\n",
    "    for i in range(0, len(org_splist)*21, len(org_splist)):  \n",
    "        with open(prepare_path + \"_mid_sp_{}_.pickle\".format(i), 'rb') as f:\n",
    "            temp_splist = pickle.load(f)\n",
    "            if mid_splist is None:\n",
    "                mid_splist = temp_splist\n",
    "            else:\n",
    "                mid_splist = mid_splist + temp_splist\n",
    "    for i in range(0, len(org_aplist)*21, len(org_aplist)):  \n",
    "        with open(prepare_path + \"_mid_ap_{}_.pickle\".format(i), 'rb') as f:\n",
    "            temp_aplist = pickle.load(f)\n",
    "            if mid_aplist is None:\n",
    "                mid_aplist = temp_aplist\n",
    "            else:\n",
    "                mid_aplist = mid_aplist + temp_aplist    \n",
    "    with open(prepare_path + \"_mid_mcep.pickle\", 'rb') as f:  \n",
    "        mid_mceplist = pickle.load(f)\n",
    "    with open(prepare_path + \"_mid_npow.pickle\", 'rb') as f:  \n",
    "        mid_npowlist = pickle.load(f)\n",
    "    with open(prepare_path + \"_mid_codeap.pickle\", 'rb') as f:  \n",
    "        mid_codeaplist = pickle.load(f) \n",
    "else:        \n",
    "    mid_f0list = []\n",
    "    mid_mceplist = []\n",
    "    mid_aplist = []\n",
    "    mid_npowlist = []\n",
    "    mid_splist = []\n",
    "    mid_codeaplist = []\n",
    "    ite = 0\n",
    "    for files in sorted(glob.iglob(pre_stored_list, recursive=True)):\n",
    "        wavf = files\n",
    "        x, fs = sf.read(wavf)\n",
    "        x = np.array(x, dtype=np.float)\n",
    "        x = low_cut_filter(x, fs, cutoff=70)\n",
    "        assert fs == 16000\n",
    "\n",
    "        print(\"extract acoustic featuers: \" + wavf)\n",
    "\n",
    "        f0, sp, ap = feat.analyze(x)\n",
    "        mcep = feat.mcep()\n",
    "        npow = feat.npow()\n",
    "        codeap = feat.codeap()\n",
    "        name, ext = os.path.splitext(wavf)\n",
    "        wav = synthesizer.synthesis_sp(f0, sp, ap)\n",
    "        wav = np.clip(wav, -32768, 32767)\n",
    "        sf.write(prepare_path + \"mid_ansys_{}_.wav\".format(ite), wav, fs)\n",
    "\n",
    "        mid_f0list.append(f0)\n",
    "        mid_splist.append(sp)\n",
    "        mid_mceplist.append(mcep)\n",
    "        mid_aplist.append(ap)\n",
    "        mid_npowlist.append(npow)\n",
    "        mid_codeaplist.append(codeap)\n",
    "\n",
    "        wav = synthesizer.synthesis(f0, mcep, ap)\n",
    "        wav = np.clip(wav, -32768, 32767)\n",
    "        sf.write(prepare_path + \"mid_mcep_{}_.wav\".format(ite), wav, fs)\n",
    "        ite = ite + 1\n",
    "        \n",
    "    with open(prepare_path + \"_mid_f0.pickle\", 'wb') as f:\n",
    "        print(f)\n",
    "        pickle.dump(mid_f0list, f)\n",
    "    with open(prepare_path + \"_mid_npow.pickle\", 'wb') as f:\n",
    "        print(f)\n",
    "        pickle.dump(mid_npowlist, f)\n",
    "    for i in range(0, len(mid_splist), len(org_splist)):\n",
    "        with open(prepare_path + \"_mid_sp_{}_.pickle\".format(i), 'wb') as f:   \n",
    "            print(f)\n",
    "            pickle.dump(mid_splist[i:i+len(org_splist)], f)\n",
    "    for i in range(0, len(mid_aplist), len(org_aplist)):\n",
    "        with open(prepare_path + \"_mid_ap_{}_.pickle\".format(i), 'wb') as f:\n",
    "            print(f)\n",
    "            pickle.dump(mid_aplist[i:i+len(org_aplist)], f)\n",
    "    with open(prepare_path + \"_mid_mcep.pickle\", 'wb') as f:\n",
    "        print(f)\n",
    "        pickle.dump(mid_mceplist, f)\n",
    "    with open(prepare_path + \"_mid_codeap.pickle\", 'wb') as f:\n",
    "        print(f)\n",
    "        pickle.dump(mid_codeaplist, f) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "class GMMTrainer(object):\n",
    "    \"\"\"\n",
    "    this class offers the training of GMM with several types of covariance matrix.\n",
    "    \n",
    "    Parameters\n",
    "    ----------\n",
    "    n_mix : int \n",
    "        the number of mixture components of the GMM\n",
    "    n_iter : int\n",
    "        the number of iteration for EM algorithm\n",
    "    covtype : str\n",
    "        the type of covariance matrix of the GMM\n",
    "        'full': full-covariance matrix\n",
    "    \n",
    "    Attributes\n",
    "    ---------\n",
    "    param : \n",
    "        sklearn-based model parameters of the GMM\n",
    "    \"\"\"\n",
    "    \n",
    "    def __init__(self, n_mix=64, n_iter=100, covtype='full', params='wmc'):\n",
    "        self.n_mix = n_mix\n",
    "        self.n_iter = n_iter\n",
    "        self.covtype = covtype\n",
    "        self.params = params\n",
    "        self.param = sklearn.mixture.GMM(n_components=self.n_mix,\n",
    "                                                     covariance_type=self.covtype,\n",
    "                                                     n_iter=self.n_iter, params=self.params)\n",
    "        \n",
    "    def train(self, jnt):\n",
    "        \"\"\"\n",
    "        fit GMM parameter from given joint feature vector\n",
    "        \n",
    "        Parametes\n",
    "        ---------\n",
    "        jnt : array, shape(`T`, `jnt.shape[0]`)\n",
    "            joint feature vector of original and target feature vector consisting of static and delta components\n",
    "        \"\"\"\n",
    "        \n",
    "        if self.covtype == 'full':\n",
    "            self.param.fit(jnt)\n",
    "            \n",
    "        return\n",
    "    \n",
    "class GMMConvertor(object):\n",
    "    \"\"\"\n",
    "    this class offers the several conversion techniques such as Maximum Likelihood Parameter Generation (MLPG)\n",
    "    and Minimum Mean Square Error (MMSE).\n",
    "    \n",
    "    Parametes\n",
    "    ---------\n",
    "    n_mix : int\n",
    "        the number of mixture components of the GMM\n",
    "    covtype : str\n",
    "        the type of covariance matrix of the GMM\n",
    "        'full': full-covariance matrix\n",
    "    gmmmode : str\n",
    "        the type of the GMM for opening\n",
    "        `None` : Normal Joint Density - GMM (JD-GMM)\n",
    "    \n",
    "    Attributes\n",
    "    ---------\n",
    "    param : \n",
    "        sklearn-based model parameters of the GMM\n",
    "    w : shape(`n_mix`)\n",
    "        vector of mixture component weight of the GMM\n",
    "    jmean : shape(`n_mix`, `jnt.shape[0]`)\n",
    "        Array of joint mean vector of the GMM\n",
    "    jcov : shape(`n_mix`, `jnt.shape[0]`, `jnt.shape[0]`)\n",
    "        array of joint covariance matrix of the GMM\n",
    "    \"\"\"\n",
    "    \n",
    "    def __init__(self, n_mix=64, covtype='full', gmmmode=None):\n",
    "        self.n_mix = n_mix\n",
    "        self.covtype = covtype\n",
    "        self.gmmmode = gmmmode\n",
    "        \n",
    "    def open_from_param(self, param):\n",
    "        \"\"\"\n",
    "        open GMM from GMMTrainer\n",
    "        \n",
    "        Parameters\n",
    "        ----------\n",
    "        param : GMMTrainer\n",
    "            GMMTrainer class\n",
    "        \"\"\"\n",
    "        \n",
    "        self.param = param\n",
    "        self._deploy_parameters()\n",
    "        \n",
    "        return\n",
    "    \n",
    "    def convert(self, data, cvtype='mlpg'):\n",
    "        \"\"\"\n",
    "        convert data based on conditional probability density function\n",
    "        \n",
    "        Parametes\n",
    "        ---------\n",
    "        data : array, shape(`T`, `dim`)\n",
    "            original data will be converted\n",
    "        cvtype : str\n",
    "            type of conversion technique\n",
    "            `mlpg` : maximum likelihood parameter generation\n",
    "            \n",
    "        Returns\n",
    "        ----------\n",
    "        odata : array, shape(`T`, `dim`)\n",
    "            converted data\n",
    "        \"\"\"\n",
    "        \n",
    "        # estimate parameter sequence\n",
    "        cseq, wseq, mseq, covseq = self._gmmmap(data)\n",
    "        \n",
    "        if cvtype == 'mlpg':\n",
    "            odata = self._mlpg(mseq, covseq)\n",
    "        else:\n",
    "            raise ValueError('please choose conversion mode in `mlpg`.')\n",
    "        \n",
    "        return odata\n",
    "    \n",
    "    def _gmmmap(self, sddata):\n",
    "        # paramete for sequencial data\n",
    "        T, sddim = sddata.shape\n",
    "        \n",
    "        # estimate posterior sequence\n",
    "        wseq = self.pX.predict_proba(sddata)\n",
    "        \n",
    "        # estimate mixture sequence\n",
    "        cseq = np.argmax(wseq, axis=1)\n",
    "        \n",
    "        mseq = np.zeros((T, sddim))\n",
    "        covseq = np.zeros((T, sddim, sddim))\n",
    "        for t in range(T):\n",
    "            # read maximum likelihood mixture component in frame t\n",
    "            m = cseq[t]\n",
    "            \n",
    "            # conditional mean vector sequence\n",
    "            mseq[t] = self.meanY[m] + self.A[m] @ (sddata[t] - self.meanX[m])\n",
    "            \n",
    "            # conditional covariance sequence\n",
    "            covseq[t] = self.cond_cov_inv[m]\n",
    "        \n",
    "        return cseq, wseq, mseq, covseq\n",
    "    \n",
    "    def _mlpg(self, mseq, covseq):\n",
    "        # parameter for sequencial data\n",
    "        T, sddim = mseq.shape\n",
    "        \n",
    "        # prepare W\n",
    "        W = construct_static_and_delta_matrix(T, sddim // 2)\n",
    "        \n",
    "        # prepare D\n",
    "        D = get_diagonal_precision_matrix(T, sddim, covseq)\n",
    "        \n",
    "        # calculate W'D\n",
    "        WD = W.T @ D\n",
    "        \n",
    "        # W'DW\n",
    "        WDW = WD @ W\n",
    "        \n",
    "        # W'Dm\n",
    "        WDM = WD @ mseq.flatten()\n",
    "        \n",
    "        # estimate y = (W'DW)^-1 * W'Dm\n",
    "        odata = scipy.sparse.linalg.spsolve(WDW, WDM, use_umfpack=False).reshape(T, sddim // 2)\n",
    "        \n",
    "        return odata\n",
    "    \n",
    "    def _deploy_parameters(self):\n",
    "        # read JD-GMM parameters from self.param\n",
    "        self.W = self.param.weights_\n",
    "        self.jmean = self.param.means_\n",
    "        self.jcov = self.param.covars_\n",
    "        \n",
    "        # devide GMM parameters into source and target parameters\n",
    "        sddim = self.jmean.shape[1] // 2\n",
    "        self.meanX = self.jmean[:, 0:sddim]\n",
    "        self.meanY = self.jmean[:, sddim:]\n",
    "        self.covXX = self.jcov[:, :sddim, :sddim]\n",
    "        self.covXY = self.jcov[:, :sddim, sddim:]\n",
    "        self.covYX = self.jcov[:, sddim:, :sddim]\n",
    "        self.covYY = self.jcov[:, sddim:, sddim:]\n",
    "        \n",
    "        # change model parameter of GMM into that of gmmmode\n",
    "        if self.gmmmode is None:\n",
    "            pass\n",
    "        else:\n",
    "            raise ValueError('please choose GMM mode in [None]')\n",
    "            \n",
    "        # estimate parameters for conversion\n",
    "        self._set_Ab()\n",
    "        self._set_pX()\n",
    "        \n",
    "        return\n",
    "    \n",
    "    def _set_Ab(self):\n",
    "        # calculate A and b from self.jmean, self.jcov\n",
    "        sddim = self.jmean.shape[1] // 2\n",
    "        \n",
    "        # calculate inverse covariance for covariance XX in each mixture\n",
    "        self.covXXinv = np.zeros((self.n_mix, sddim, sddim))\n",
    "        for m in range(self.n_mix):\n",
    "            self.covXXinv[m] = np.linalg.inv(self.covXX[m])\n",
    "            \n",
    "        # calculate A, b, and conditional covariance given X\n",
    "        self.A = np.zeros((self.n_mix, sddim, sddim))\n",
    "        self.b = np.zeros((self.n_mix, sddim))\n",
    "        self.cond_cov_inv = np.zeros((self.n_mix, sddim, sddim))\n",
    "        for m in range(self.n_mix):\n",
    "            # calculate A (A = yxcov_m * xxcov_m^-1)\n",
    "            self.A[m] = self.covYX[m] @ self.covXXinv[m]\n",
    "            \n",
    "            # calculate b (b = mean^Y - A * mean^X)\n",
    "            self.b[m] = self.meanY[m] - self.A[m] @ self.meanX[m]\n",
    "            \n",
    "            # calculate conditional covariance (cov^(Y|X)^-1 = (yycov - A * xycov)^-1)\n",
    "            self.cond_cov_inv[m] = np.linalg.inv(self.covYY[m] - self.A[m] @ self.covXY[m])\n",
    "            \n",
    "        return\n",
    "    \n",
    "    def _set_pX(self):\n",
    "        # probability density function of X \n",
    "        self.pX = sklearn.mixture.GMM(n_components=self.n_mix, covariance_type=self.covtype)\n",
    "        self.pX.weights_ = self.W\n",
    "        self.pX.means_ = self.meanX\n",
    "        self.pX.covars_ = self.covXX\n",
    "        \n",
    "        # following function is required to estimate porsterior\n",
    "        # p(x | \\lambda^(X))\n",
    "        #self.pX.precisions_cholesky_ = _compute_precision_cholesky(self.covXX, self.covtype)\n",
    "        \n",
    "        return\n",
    "    \n",
    "def get_diagonal_precision_matrix(T, D, covseq):\n",
    "    return scipy.sparse.block_diag(covseq, format='csr')        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_alignment(odata, onpow, tdata, tnpow, opow=-20, tpow=-20, sd=0, cvdata=None, given_twf=None, otflag=None, distance='melcd'):\n",
    "    \"\"\"\n",
    "    get alignment between original and target.\n",
    "    \n",
    "    Parameters\n",
    "    ----------\n",
    "    odata : array, shape(`T`, `dim`)\n",
    "        acoustic feature vector of original\n",
    "    onpow : array, shape(`T`)\n",
    "        Normalized power vector of original\n",
    "    tdata : array, shape(`T`, `dim`)\n",
    "        acoustic feature vector of target\n",
    "    tnpow : array, shape(`T`)\n",
    "        Normalized power vector of target\n",
    "    opow : float\n",
    "        power threshold of original\n",
    "    tpow : float\n",
    "        power threshold of target\n",
    "    sd : int\n",
    "        start dimension to be used for alignment\n",
    "    cvdata : array, shape(`T`, `dim`)\n",
    "        converted original data\n",
    "    given_twf : array, shape(`T_new`, `dim * 2`)\n",
    "        Alignment given twf\n",
    "    otflag : str\n",
    "        Alignment into the length of specification\n",
    "        'org' : alignment into original length\n",
    "        'tar' : alignment into target length\n",
    "    distance : str\n",
    "        Distance function to be used\n",
    "    \n",
    "    Returns\n",
    "    ----------\n",
    "    jdata : array, shape(`T_new`, `dim * 2`)\n",
    "        joint static and delta feature vector\n",
    "    twf : array, shape(`T_new`, `dim * 2`)\n",
    "        Time warping function\n",
    "    mcd : float\n",
    "        Mel-cepstrum distortion between arrays\n",
    "    \"\"\"\n",
    "    \n",
    "    oexdata = extsddata(odata[:, sd:], onpow, power_threshold=opow)\n",
    "    texdata = extsddata(tdata[:, sd:], tnpow, power_threshold=tpow)\n",
    "    \n",
    "    if cvdata is None:\n",
    "        align_odata = oexdata\n",
    "    else:\n",
    "        cvexdata = extsddata(cvdata, onpow, power_threshold=opow)\n",
    "        align_odata = cvexdata\n",
    "    \n",
    "    if given_twf is None: \n",
    "        twf = estimate_twf(align_odata, texdata, distance=distance, fast=False, otflag=otflag)\n",
    "    else:\n",
    "        twf = given_twf\n",
    "    \n",
    "    jdata = align_data(oexdata, texdata, twf)\n",
    "    mcd = melcd(align_odata[twf[0]], texdata[twf[1]])\n",
    "    \n",
    "    return jdata, twf, mcd\n",
    "\n",
    "def align_feature_vectors(odata, onpows, tdata, tnpows, opow=-100, tpow=-100, itnum=3, sd=0, given_twfs=None, otflag=None):\n",
    "    \"\"\"\n",
    "    get alignment to create joint feature vector\n",
    "    \n",
    "    Parameters\n",
    "    ----------\n",
    "    odata : list, (`num_files`)\n",
    "        List of original feature vectors\n",
    "    onpow : list, (`num_files`)\n",
    "        List of original npows\n",
    "    tdata : list, (`num_files`)\n",
    "        List of target feature vectors\n",
    "    tnpow : list, (`num_files`)\n",
    "        List of target npows\n",
    "    opow : float\n",
    "        power threshold of original\n",
    "    tpow : float\n",
    "        power threshold of target\n",
    "    itnum : int\n",
    "        the number of iteration\n",
    "    sd : int\n",
    "        start dimension of feature vector to be used for alignment\n",
    "    given_twf : array, shape(`T_new`, `dim * 2`)\n",
    "        use given alignment while 1st iteration\n",
    "    otflag : str\n",
    "        Alignment into the length of specification\n",
    "        'org' : alignment into original length\n",
    "        'tar' : alignment into target length\n",
    "    distance : str\n",
    "        Distance function to be used\n",
    "    \n",
    "    Returns\n",
    "    ----------\n",
    "    jdata : array, shape(`T_new`, `dim * 2`)\n",
    "        joint static and delta feature vector\n",
    "    twf : array, shape(`T_new`, `dim * 2`)\n",
    "        Time warping function\n",
    "    mcd : float\n",
    "        Mel-cepstrum distortion between arrays\n",
    "    \"\"\"\n",
    "    \n",
    "    it = 1\n",
    "    num_files = len(odata)\n",
    "    cvgmm, cvdata = None, None\n",
    "    for it in range(1, itnum+1):\n",
    "        print('{}-th joint feature extraction starts.'.format(it))\n",
    "        \n",
    "        # alignment\n",
    "        twfs, jfvs = [], []\n",
    "        for i in range(num_files):\n",
    "            if it == 1 and given_twfs is not None:\n",
    "                gtwf = given_twfs[i]\n",
    "            else:\n",
    "                gtwf = None\n",
    "            \n",
    "            if it > 1:\n",
    "                cvdata = cvgmm.convert(static_delta(odata[i][:, sd:]))\n",
    "            \n",
    "            jdata, twf, mcd = get_alignment(odata[i], onpows[i], tdata[i], tnpows[i], opow=opow, tpow=tpow,\n",
    "                                            sd=sd, cvdata=cvdata, given_twf=gtwf, otflag=otflag)\n",
    "            twfs.append(twf)\n",
    "            jfvs.append(jdata)\n",
    "            print('distortion [dB] for {}-th file: {}'.format(i+1, mcd))\n",
    "            \n",
    "        jnt_data = transform_jnt(jfvs)\n",
    "        \n",
    "        if it != itnum:\n",
    "            # train GMM, if not final iteration\n",
    "            datagmm = GMMTrainer()\n",
    "            datagmm.train(jnt_data)\n",
    "            cvgmm = GMMConvertor()\n",
    "            cvgmm.open_from_param(datagmm.param)\n",
    "        it += 1\n",
    "    return jfvs, twfs    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 2. estimate twf and jnt\n",
    "if os.path.exists(prepare_path + \"_jnt_mcep_0_.pickle\"):\n",
    "    pass\n",
    "else:\n",
    "    for i in range(0, len(mid_mceplist), len(org_mceplist)):\n",
    "        org_mceps = org_mceplist\n",
    "        org_npows = org_npowlist\n",
    "        mid_mceps = mid_mceplist[i:i+len(org_mceps)]\n",
    "        mid_npows = mid_npowlist[i:i+len(org_npows)]\n",
    "        assert len(org_mceps) == len(mid_mceps)\n",
    "        assert len(org_npows) == len(mid_npows)\n",
    "        assert len(org_mceps) == len(org_npows)\n",
    "\n",
    "        # dtw between original and target 0-th and silence\n",
    "        print(\"## alignment mcep 0-th and silence ##\")\n",
    "        jmceps, twfs = align_feature_vectors(org_mceps, org_npows, mid_mceps, mid_npows, opow=-15, tpow=-15, sd=1)\n",
    "        jnt_mcep = transform_jnt(jmceps)\n",
    "\n",
    "        # save joint feature vectors\n",
    "        with open(prepare_path + \"_jnt_mcep_{}_.pickle\".format(i), 'wb') as f:   \n",
    "            print(f)\n",
    "            pickle.dump(jnt_mcep, f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<_io.BufferedReader name='./utterance/pre-stored-en/output/initgmm.pickle'>\n"
     ]
    }
   ],
   "source": [
    "# 3. make EV-GMM\n",
    "initgmm, initgmm_codeap = None, None\n",
    "if os.path.exists(prepare_path + \"initgmm.pickle\"):\n",
    "    with open(prepare_path + \"initgmm.pickle\".format(i), 'rb') as f:   \n",
    "        print(f)\n",
    "        initgmm = pickle.load(f)\n",
    "else:\n",
    "    jnt, jnt_codeap = None, []\n",
    "    for i in range(0, len(mid_mceplist), len(org_mceplist)):\n",
    "        with open(prepare_path + \"_jnt_mcep_{}_.pickle\".format(i), 'rb') as f:  \n",
    "            temp_jnt = pickle.load(f)\n",
    "            if jnt is None:\n",
    "                jnt = temp_jnt\n",
    "            else:\n",
    "                jnt = np.r_[jnt, temp_jnt]\n",
    "\n",
    "    # train initial gmm\n",
    "    initgmm = GMMTrainer()\n",
    "    initgmm.train(jnt)\n",
    "\n",
    "    with open(prepare_path + \"initgmm.pickle\", 'wb') as f:   \n",
    "        print(f)\n",
    "        pickle.dump(initgmm, f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "# get initial gmm params\n",
    "init_W = initgmm.param.weights_\n",
    "init_jmean = initgmm.param.means_\n",
    "init_jcov = initgmm.param.covars_\n",
    "sddim = init_jmean.shape[1] // 2\n",
    "init_meanX = init_jmean[:, :sddim]\n",
    "init_meanY = init_jmean[:, sddim:]\n",
    "init_covXX = init_jcov[:, :sddim, :sddim]\n",
    "init_covXY = init_jcov[:, :sddim, sddim:]\n",
    "init_covYX = init_jcov[:, sddim:, :sddim]\n",
    "init_covYY = init_jcov[:, sddim:, sddim:]\n",
    "fitted_source = init_meanX\n",
    "fitted_target = init_meanY\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "sv = None\n",
    "if os.path.exists(prepare_path + \"_sv.npy\"):\n",
    "    sv = np.array(sv)\n",
    "    sv = np.load(prepare_path + '_sv.npy')\n",
    "else:\n",
    "    depengmm, depenjnt = None, None\n",
    "    sv = []\n",
    "    for i in range(0, len(mid_mceplist), len(org_mceplist)):\n",
    "        with open(prepare_path + \"_jnt_mcep_{}_.pickle\".format(i), 'rb') as f:  \n",
    "            depenjnt = pickle.load(f)\n",
    "            depengmm = GMMTrainer(params='m')\n",
    "            depengmm.param.weights_ = init_W\n",
    "            depengmm.param.means_ = init_jmean\n",
    "            depengmm.param.covars_ = init_jcov\n",
    "            depengmm.train(depenjnt)\n",
    "            sv.append(depengmm.param.means_)\n",
    "    sv = np.array(sv)\n",
    "    np.save(prepare_path + \"_sv\", sv)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "n_mix = 64\n",
    "S = int(len(mid_mceplist) / len(org_mceplist))\n",
    "assert S == 22\n",
    "\n",
    "source_pca = sklearn.decomposition.PCA()\n",
    "source_pca.fit(sv[:,:,:sddim].reshape((S, n_mix*sddim)))\n",
    "\n",
    "target_pca = sklearn.decomposition.PCA()\n",
    "target_pca.fit(sv[:,:,sddim:].reshape((S, n_mix*sddim)))\n",
    "\n",
    "eigenvectors = source_pca.components_.reshape((n_mix, sddim, S)), target_pca.components_.reshape((n_mix, sddim, S)) \n",
    "biasvectors = source_pca.mean_.reshape((n_mix, sddim)), target_pca.mean_.reshape((n_mix, sddim))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A11.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A14.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A17.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A18.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A19.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A20.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A21.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A22.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A23.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A24.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A25.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A26.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A27.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A28.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A29.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A30.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A31.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A32.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A33.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A34.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A35.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A36.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A37.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A38.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A39.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A40.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A41.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A42.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A43.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A44.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A45.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A46.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A47.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A48.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A49.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A50.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A51.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A52.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A53.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A54.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A55.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A56.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A57.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A58.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/input/EJM10/V01/T01/TIMIT/000/A59.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/ATR503/A/A01.wav\n"
     ]
    }
   ],
   "source": [
    "# estimate statistic features\n",
    "for_convert_source = __same_path + 'input/EJM10/V01/T01/TIMIT/000/*.wav'\n",
    "for_convert_target = __same_path + 'adaptation/EJM04/V01/T01/ATR503/A/*.wav'\n",
    "\n",
    "src_f0list = []\n",
    "src_splist = []\n",
    "src_mceplist = []\n",
    "src_aplist = []\n",
    "src_npowlist = []\n",
    "src_codeaplist = []\n",
    "ite = 0\n",
    "for files in sorted(glob.iglob(for_convert_source, recursive=True)):\n",
    "    wavf = files\n",
    "    x, fs = sf.read(wavf)\n",
    "    x = np.array(x, dtype=np.float)\n",
    "    x = low_cut_filter(x, fs, cutoff=70)\n",
    "    assert fs == 16000\n",
    "\n",
    "    print(\"extract acoustic featuers: \" + wavf)\n",
    "\n",
    "    f0, sp, ap = feat.analyze(x)\n",
    "    mcep = feat.mcep()\n",
    "    npow = feat.npow()\n",
    "    codeap = feat.codeap()\n",
    "    wav = synthesizer.synthesis_sp(f0, sp, ap)\n",
    "    wav = np.clip(wav, -32768, 32767)\n",
    "    sf.write(output_path + \"input_ansys_{}_.wav\".format(ite), wav, fs)\n",
    "\n",
    "    src_f0list.append(f0)\n",
    "    src_splist.append(sp)\n",
    "    src_mceplist.append(mcep)\n",
    "    src_aplist.append(ap)\n",
    "    src_npowlist.append(npow)\n",
    "    src_codeaplist.append(codeap)\n",
    "\n",
    "    wav = synthesizer.synthesis(f0, mcep, ap)\n",
    "    wav = np.clip(wav, -32768, 32767)\n",
    "    sf.write(output_path + \"input_mcep_{}_.wav\".format(ite), wav, fs)\n",
    "    ite = ite + 1\n",
    "\n",
    "tar_f0list = []\n",
    "tar_mceplist = []\n",
    "tar_aplist = []\n",
    "tar_npowlist = []\n",
    "tar_splist = []\n",
    "tar_codeaplist = []\n",
    "ite = 0\n",
    "for files in sorted(glob.iglob(for_convert_target, recursive=True)):\n",
    "    wavf = files\n",
    "    x, fs = sf.read(wavf)\n",
    "    x = np.array(x, dtype=np.float)\n",
    "    x = low_cut_filter(x, fs, cutoff=70)\n",
    "    assert fs == 16000\n",
    "\n",
    "    print(\"extract acoustic featuers: \" + wavf)\n",
    "\n",
    "    f0, sp, ap = feat.analyze(x)\n",
    "    mcep = feat.mcep()\n",
    "    npow = feat.npow()\n",
    "    codeap = feat.codeap()\n",
    "    name, ext = os.path.splitext(wavf)\n",
    "    wav = synthesizer.synthesis_sp(f0, sp, ap)\n",
    "    wav = np.clip(wav, -32768, 32767)\n",
    "    sf.write(output_path + \"target_ansys_{}_.wav\".format(ite), wav, fs)\n",
    "\n",
    "    tar_f0list.append(f0)\n",
    "    tar_splist.append(sp)\n",
    "    tar_mceplist.append(mcep)\n",
    "    tar_aplist.append(ap)\n",
    "    tar_npowlist.append(npow)\n",
    "    tar_codeaplist.append(codeap)\n",
    "\n",
    "    wav = synthesizer.synthesis(f0, mcep, ap)\n",
    "    wav = np.clip(wav, -32768, 32767)\n",
    "    sf.write(output_path + \"target_mcep_{}_.wav\".format(ite), wav, fs)\n",
    "    ite = ite + 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "f0stats = F0statistics()\n",
    "srcf0stats = f0stats.estimate(org_f0list)\n",
    "tarf0stats = f0stats.estimate(tar_f0list)\n",
    "\n",
    "gv = GV()\n",
    "srcgvstats = gv.estimate(org_mceplist)\n",
    "targvstats = gv.estimate(tar_mceplist)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "adapt:  1 / 1\n",
      "epoch =  0\n",
      "epoch =  1\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  2\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  3\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  4\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  5\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  6\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  7\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  8\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  9\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  10\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  11\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  12\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  13\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  14\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  15\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  16\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  17\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  18\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  19\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  20\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  21\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  22\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  23\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  24\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  25\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  26\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  27\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  28\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  29\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  30\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  31\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  32\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  33\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  34\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  35\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  36\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  37\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  38\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  39\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  40\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  41\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  42\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  43\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  44\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  45\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  46\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  47\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  48\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  49\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  50\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  51\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  52\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  53\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  54\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  55\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  56\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  57\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  58\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  59\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  60\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  61\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  62\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  63\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  64\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  65\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  66\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  67\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  68\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  69\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  70\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  71\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  72\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  73\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  74\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  75\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  76\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  77\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  78\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  79\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  80\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  81\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  82\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  83\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  84\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  85\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  86\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  87\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  88\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  89\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  90\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  91\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  92\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  93\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  94\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  95\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  96\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  97\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  98\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch =  99\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    }
   ],
   "source": [
    "# 5. fitting target\n",
    "epoch = 100\n",
    "\n",
    "fitgmm = sklearn.mixture.GMM(n_components=n_mix,\n",
    "                             covariance_type='full',\n",
    "                             n_iter=100)\n",
    "fitgmm.weights_ = init_W\n",
    "fitgmm.means_ = init_meanY\n",
    "fitgmm.covars_ = init_covYY\n",
    "\n",
    "for i in range(len(tar_mceplist)):\n",
    "    print(\"adapt: \", i+1, \"/\", len(tar_mceplist))\n",
    "    target = tar_mceplist[i]\n",
    "    target_pow = target[:, 0]\n",
    "    target = target[:, 1:]\n",
    "    for x in range(epoch):\n",
    "        print(\"epoch = \", x)\n",
    "        predict = fitgmm.predict_proba(np.atleast_2d(static_delta(target)))\n",
    "        \n",
    "        y = np.sum([predict[:, k:k+1] * (static_delta(target) - biasvectors[1][k]) for k in range(n_mix)], axis=1)\n",
    "        gamma = np.sum(predict, axis=0)\n",
    "        \n",
    "        left = np.sum([gamma[k] * np.dot(eigenvectors[1][k].T,\n",
    "                                         np.linalg.solve(fitgmm.covars_, eigenvectors[1])[k])\n",
    "                                                        for k in range(n_mix)], axis=0)\n",
    "        right = np.sum([np.dot(eigenvectors[1][k].T,\n",
    "                               np.linalg.solve(fitgmm.covars_, y)[k])\n",
    "                        for k in range(n_mix)], axis=0)\n",
    "        weight = np.linalg.solve(left, right)\n",
    "        \n",
    "        fitted_target = np.dot(eigenvectors[1], weight) + biasvectors[1]\n",
    "        fitgmm.means_ = fitted_target"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "def mcepconvert(source, weights, jmean, meanX, covarXX, covarXY, covarYX, covarYY,\n",
    "                fitted_source, fitted_target):\n",
    "    \n",
    "    M = 64\n",
    "    # set pX\n",
    "    px = sklearn.mixture.GMM(n_components=M, covariance_type='full', n_iter=100)\n",
    "    px.weights_ = weights\n",
    "    px.means_ = meanX\n",
    "    px.covars_ = covarXX\n",
    "    \n",
    "    # set Ab\n",
    "    sddim = jmean.shape[1] // 2\n",
    "    covXXinv = np.zeros((M, sddim, sddim))\n",
    "    for m in range(M):\n",
    "        covXXinv[m] = np.linalg.inv(covarXX[m])\n",
    "    A = np.zeros((M, sddim, sddim))\n",
    "    b = np.zeros((M, sddim))\n",
    "    cond_cov_inv = np.zeros((M, sddim, sddim))\n",
    "    for m in range(M):\n",
    "        A[m] = covarYX[m] @ covXXinv[m]\n",
    "        b[m] = fitted_target[m] - A[m] @ meanX[m]\n",
    "        cond_cov_inv[m] = np.linalg.inv(covarYY[m] - A[m] @ covarXY[m])\n",
    "        \n",
    "    # _gmmmap\n",
    "    T, sddim = source.shape\n",
    "    wseq = px.predict_proba(source)\n",
    "    cseq = np.argmax(wseq, axis=1)\n",
    "    mseq = np.zeros((T, sddim))\n",
    "    covseq = np.zeros((T, sddim, sddim))\n",
    "    for t in range(T):\n",
    "        m = cseq[t]\n",
    "        mseq[t] = fitted_target[m] + A[m] @ (source[t] - meanX[m])\n",
    "        covseq[t] = cond_cov_inv[m]\n",
    "        \n",
    "    # _mlpg\n",
    "    T, sddim = mseq.shape\n",
    "    W = construct_static_and_delta_matrix(T, sddim // 2)\n",
    "    D = get_diagonal_precision_matrix(T, sddim, covseq)\n",
    "    WD = W.T @ D\n",
    "    WDW = WD @ W\n",
    "    WDM = WD @ mseq.flatten()\n",
    "    \n",
    "    output = scipy.sparse.linalg.spsolve(WDW, WDM, use_umfpack=False).reshape(T, sddim // 2)\n",
    "    return output"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:57: DeprecationWarning: Class GMM is deprecated; The class GMM is deprecated in 0.18 and will be  removed in 0.20. Use class GaussianMixture instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "/home/higa/anaconda3/lib/python3.6/site-packages/sklearn/utils/deprecation.py:75: DeprecationWarning: Function log_multivariate_normal_density is deprecated; The function log_multivariate_normal_density is deprecated in 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    }
   ],
   "source": [
    "# learn cvgvstats \n",
    "cv_mceps = []\n",
    "for i in range(len(src_mceplist)):\n",
    "    temp_mcep = src_mceplist[i]\n",
    "    temp_mcep_0th = temp_mcep[:, 0]\n",
    "    temp_mcep = temp_mcep[:, 1:]\n",
    "    sta_mcep = static_delta(temp_mcep)\n",
    "    cvmcep_wopow = np.array(mcepconvert(sta_mcep, init_W, init_jmean, init_meanX,\n",
    "                                        init_covXX, init_covXY, init_covYX, init_covYY,\n",
    "                                        fitted_source, fitted_target))\n",
    "    cvmcep = np.c_[temp_mcep_0th, cvmcep_wopow]\n",
    "    cv_mceps.append(cvmcep)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "cvgvstats = gv.estimate(cv_mceps)\n",
    "\n",
    "for i in range(len(src_mceplist)):\n",
    "    cvmcep_wGV = gv.postfilter(cv_mceps[i], targvstats, cvgvstats=cvgvstats)\n",
    "    cvf0 = f0stats.convert(src_f0list[i], srcf0stats, tarf0stats)\n",
    "    wav = synthesizer.synthesis(cvf0, cvmcep_wGV, src_aplist[i], rmcep=src_mceplist[i])\n",
    "    sf.write(output_path + \"cv_{}_.wav\".format(i), wav, 16000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "for i in range(len(src_mceplist)):\n",
    "    wav = synthesizer.synthesis(src_f0list[i], src_mceplist[i], src_aplist[i])\n",
    "    sf.write(output_path + \"mcep_{}_.wav\".format(i), wav, 16000)\n",
    "    wav = synthesizer.synthesis_sp(src_f0list[i], src_splist[i], src_aplist[i])\n",
    "    sf.write(output_path + \"ansys_{}_.wav\".format(i), wav, 16000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x7f004c43b5f8>]"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA6MAAAEvCAYAAACwixeWAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRzIKAAAAALDdfwG69KyMutdzmAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f004b36ea20>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "cvf0 = f0stats.convert(src_f0list[0], srcf0stats, tarf0stats)\n",
    "plt.plot(cvf0)\n",
    "#plt.plot(src_f0list[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "cvmcep_wGV = gv.postfilter(cv_mceps[0], srcgvstats, cvgvstats=cvgvstats)\n",
    "cvf0 = f0stats.convert(src_f0list[0], srcf0stats, tarf0stats)\n",
    "wav = synthesizer.synthesis(cvf0, cvmcep_wGV, src_aplist[0], rmcep=src_mceplist[0])\n",
    "sf.write(output_path + \"te.wav\", wav, 16000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A01.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A02.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A03.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A04.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A05.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A06.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A07.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A08.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A09.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A10.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A11.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A12.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A13.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A14.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A15.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A16.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A17.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A18.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A19.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A20.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A21.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A22.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A23.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A24.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A25.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A26.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A27.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A28.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A29.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A30.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A31.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A32.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A33.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A34.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A35.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A36.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A37.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A38.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A39.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A40.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A41.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A42.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A43.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A44.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A45.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A46.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A47.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A48.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A49.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A50.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A51.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A52.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A53.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A54.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A55.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A56.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A57.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A58.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A59.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A60.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A61.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A62.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A63.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A64.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A65.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A66.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A67.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A68.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A69.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A70.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A71.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A72.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A73.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A74.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A75.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A76.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A77.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A78.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A79.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A80.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A81.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A82.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A83.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A84.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A85.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A86.wav\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A87.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A88.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A89.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A90.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A91.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A92.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A93.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A94.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A95.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A96.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A97.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A98.wav\n",
      "extract acoustic featuers: ./utterance/pre-stored-en/adaptation/EJM04/V01/T01/TIMIT/000/A99.wav\n"
     ]
    }
   ],
   "source": [
    "# org-cv distance\n",
    "with open(output_path + \"melcd_org-cv.txt\", \"w\") as outfile:\n",
    "    outfile.write(\"adapt1 org-cv mcd.\\n\")\n",
    "    for i in range(len(src_mceplist)):\n",
    "        temp_mcep = src_mceplist[i]\n",
    "        temp_mcep_0th = temp_mcep[:, 0]\n",
    "        temp_mcep = temp_mcep[:, 1:]\n",
    "        \n",
    "        temp_cv = cv_mceps[i]\n",
    "        temp_cv_0th = temp_cv[:, 0]\n",
    "        temp_cv = temp_cv[:, 1:]\n",
    "        \n",
    "        _, _, mcd = get_alignment(temp_mcep, temp_mcep_0th, temp_cv, temp_cv_0th, opow=-15, tpow=-15, sd=1)\n",
    "        outfile.write(\"{0},{1}\\n\".format(i, mcd))\n",
    "\n",
    "# cv-target distance\n",
    "# read target files and analyze mceps\n",
    "targets_mceplist = []\n",
    "targets_list = __same_path + 'adaptation/EJM04/V01/T01/TIMIT/000/*.wav'\n",
    "for files in sorted(glob.iglob(targets_list, recursive=True)):\n",
    "    wavf = files\n",
    "    x, fs = sf.read(wavf)\n",
    "    x = np.array(x, dtype=np.float)\n",
    "    x = low_cut_filter(x, fs, cutoff=70)\n",
    "    assert fs == 16000\n",
    "    print(\"extract acoustic featuers: \" + wavf)\n",
    "\n",
    "    f0, sp, ap = feat.analyze(x)\n",
    "    mcep = feat.mcep()\n",
    "    targets_mceplist.append(mcep)\n",
    "\n",
    "with open(output_path + \"melcd_cv-target.txt\", \"w\") as outfile:\n",
    "    outfile.write(\"adapt1 cv-target mcd.\\n\")\n",
    "    \n",
    "    for i in range(len(src_mceplist)):\n",
    "        temp_mcep = targets_mceplist[i]\n",
    "        temp_mcep_0th = temp_mcep[:, 0]\n",
    "        temp_mcep = temp_mcep[:, 1:]\n",
    "        \n",
    "        temp_cv = cv_mceps[i]\n",
    "        temp_cv_0th = temp_cv[:, 0]\n",
    "        temp_cv = temp_cv[:, 1:]\n",
    "        \n",
    "        _, _, mcd = get_alignment(temp_cv, temp_cv_0th, temp_mcep, temp_mcep_0th, opow=-15, tpow=-15, sd=1)\n",
    "        outfile.write(\"{0},{1}\\n\".format(i, mcd))\n",
    "\n",
    "# org-target distance\n",
    "with open(output_path + \"melcd_org-target.txt\", \"w\") as outfile:\n",
    "    outfile.write(\"adapt1 org-target mcd.\\n\")\n",
    "    \n",
    "    for i in range(len(src_mceplist)):\n",
    "\n",
    "        temp_mcep = src_mceplist[i]\n",
    "        temp_mcep_0th = temp_mcep[:, 0]\n",
    "        temp_mcep = temp_mcep[:, 1:]\n",
    "        \n",
    "        temp_mcep2 = targets_mceplist[i]\n",
    "        temp_mcep2_0th = temp_mcep2[:, 0]\n",
    "        temp_mcep2 = temp_mcep2[:, 1:]\n",
    "        \n",
    "        \n",
    "        _, _, mcd = get_alignment(temp_mcep, temp_mcep_0th, temp_mcep2, temp_mcep2_0th, opow=-15, tpow=-15, sd=1)\n",
    "        outfile.write(\"{0},{1}\\n\".format(i, mcd))"
   ]
  }
 ],
 "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
}