【python_speech_features】MFCC调参用法以及参数说明

代码调用

from python_speech_features import mfcc
mfcc_feature = mfcc(**kwargs)

params

• signal:
  the audio signal from which to compute features.Should be an N1 array
  用来计算梅尔频率倒谱系数特性的音频信号。是一个N1的数组
• samplerate:
  the samplerate of the signal we are working with.
  音频信号的采样率
• winlen:
  the length of the analysis window in seconds. Default is 0.025s (25 milliseconds)
  分析窗口的长度，以秒为单位。默认值为0.025s(25毫秒)（ps: 短时傅里叶变换的窗口长度）
• winstep:
  the step between successive windows in seconds. Default is 0.01s (10 milliseconds)
  连续窗口之间的步长，以秒为单位。默认值为0.01s(10毫秒)
• numcep:
  the number of cepstrum to return, default 13
  返回倒谱的数量，默认为13
• nfilt:
  the number of filters in the filterbank, default 26.
  滤波器组中的过滤器数量，默认为26个。
• nfft:
  the FFT size. Default is 512.
  FFT大小。默认是512。
• lowfreq:
  lowest band edge of mel filters. In Hz, default is 0.
  梅尔滤波器的最低频带边缘。在频率(HZ)中，默认值为0。
• highfreq:
  highest band edge of mel filters. In Hz, default is samplerate/2
  梅尔过滤器的最高频带边缘。在频率(HZ)中，默认值为1/2倍音频信号采样率
• preemph:
  apply preemphasis filter with preemph as coefficient. 0 is no filter. Default is 0.97.
  采用preemph为系数的预加重滤波器。0不是过滤器。默认是0.97。
• ceplifter:
  apply a lifter to final cepstral coefficients. 0 is no lifter. Default is 22.
  将一个lifter应用到最终倒谱系数。0不是lifter。默认是22。
• appendEnergy:
  if this is true, the zeroth cepstral coefficient is replaced with the log of the total frame energy.
  如果这个参数的值是True，第0阶倒谱系数被替换为总帧能量的对数。
• winfunc:
  the analysis window to apply to each frame. By default no window is applied. You can use numpy window functions here e.g. winfunc=numpy.hamming
  分析窗口应用于每一帧。默认情况下不应用任何窗口。您可以在这里使用numpy窗口函数，例如winfunc=numpy.hamming(汉明窗)c

Return

• A numpy array of size (NUMFRAMES by numcep) containing
  features. Each row holds 1 feature vector.
  返回一个大小(NUMFRAMES窗口数量 , numcep倒谱数量)的numpy数组，其中包含一些特性。每一行包含一个特征向量。