Source Code for Module timeside.analyzer.irit_speech_entropy

  1  # -*- coding: utf-8 -*- 
  2  # 
  3  # Copyright (c) 2013 Maxime Le Coz <lecoz@irit.fr> 
  4   
  5  # This file is part of TimeSide. 
  6   
  7  # TimeSide is free software: you can redistribute it and/or modify 
  8  # it under the terms of the GNU General Public License as published by 
  9  # the Free Software Foundation, either version 2 of the License, or 
 10  # (at your option) any later version. 
 11   
 12  # TimeSide is distributed in the hope that it will be useful, 
 13  # but WITHOUT ANY WARRANTY; without even the implied warranty of 
 14  # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the 
 15  # GNU General Public License for more details. 
 16   
 17  # You should have received a copy of the GNU General Public License 
 18  # along with TimeSide.  If not, see <http://www.gnu.org/licenses/>. 
 19   
 20  # Author: Maxime Le Coz <lecoz@irit.fr> 
 21   
 22  from timeside.core import Processor, implements, interfacedoc 
 23  from timeside.analyzer.core import Analyzer 
 24  from timeside.analyzer.utils import entropy, computeModulation 
 25  from timeside.analyzer.utils import segmentFromValues 
 26  from timeside.api import IAnalyzer 
 27  from numpy import array 
 28  from scipy.ndimage.morphology import binary_opening 
 29   
 30   
 31 -class IRITSpeechEntropy(Analyzer): 
 32      implements(IAnalyzer) 
 33   
 34      @interfacedoc 
 35 -    def setup(self, channels=None, samplerate=None, blocksize=None, totalframes=None): 
 36          super(IRITSpeechEntropy, self).setup( 
 37              channels, samplerate, blocksize, totalframes) 
 38          self.entropyValue = [] 
 39          self.threshold = 0.4 
 40          self.smoothLen = 5 
 41          self.modulLen = 2 
 42   
 43      @staticmethod 
 44      @interfacedoc 
 45 -    def id(): 
 46          return "irit_speech_entropy" 
 47   
 48      @staticmethod 
 49      @interfacedoc 
 50 -    def name(): 
 51          return "IRIT Speech entropy" 
 52   
 53      @staticmethod 
 54      @interfacedoc 
 55 -    def unit(): 
 56          return "" 
 57   
 58 -    def __str__(self): 
 59          return "Speech confidences indexes" 
 60   
 61 -    def process(self, frames, eod=False): 
 62          self.entropyValue.append(entropy(frames)) 
 63          return frames, eod 
 64   
 65 -    def post_process(self): 
 66   
 67          entropyValue = array(self.entropyValue) 
 68          w = self.modulLen * self.samplerate() / self.blocksize() 
 69          modulentropy = computeModulation(entropyValue, w, False) 
 70          confEntropy = array(modulentropy - self.threshold) / self.threshold 
 71          confEntropy[confEntropy > 1] = 1 
 72   
 73          conf = self.new_result(data_mode='value', time_mode='framewise') 
 74   
 75          conf.id_metadata.id += '.' + 'confidence' 
 76          conf.id_metadata.name += ' ' + 'Confidence' 
 77   
 78          conf.data_object.value = confEntropy 
 79          self._results.add(conf) 
 80   
 81          # Binary Entropy 
 82          binaryEntropy = modulentropy > self.threshold 
 83          binaryEntropy = binary_opening( 
 84              binaryEntropy, [1] * (self.smoothLen * 2)) 
 85   
 86          convert = {False: 0, True: 1} 
 87          label = {0: 'NonSpeech', 1: 'Speech'} 
 88          segList = segmentFromValues(binaryEntropy) 
 89   
 90   
 91   
 92          segs = self.new_result(data_mode='label', time_mode='segment') 
 93          segs.id_metadata.id += '.' + 'segments' 
 94          segs.id_metadata.name += ' ' + 'Segments' 
 95   
 96          segs.data_object.label = segList 
 97   
 98          segs.data_object.label = [convert[s[2]] for s in segList] 
 99          segs.data_object.time = [(float(s[0]) * self.blocksize() / 
100                                    self.samplerate()) 
101                                    for s in segList] 
102          segs.data_object.duration = [(float(s[1]-s[0]) * self.blocksize() / 
103                                    self.samplerate()) 
104                                    for s in segList] 
105   
106          self._results.add(segs) 
107   
108          return 
109