语音识别(4)-- 提取频域特征_基于频域特征提取的语音识别

提取频域特征

# -*- coding:utf-8 -*-
'''
  提取频域特征 --使用梅尔频率倒谱系数
'''
import numpy as np
import matplotlib.pyplot as plt
from scipy.io import wavfile
from python_speech_features import mfcc,logfbank

# 读取输入音频文件
sampling_freq,audio=wavfile.read('input_freq.wav')
# 提取mfcc和过滤器特征
mfcc_features=mfcc(audio,sampling_freq)
filter_features=logfbank(audio,sampling_freq)
# 打印参数
print '\nMFCC :\n Number of window =',mfcc_features.shape[0]
print '每个特征的长度为: ',mfcc_features.shape[1]
print '\n Filter bank: \n Number of window =',filter_features.shape[0]
print '每个特征的长度为: ',filter_features.shape[1]
# 画出特征图
mfcc_features=mfcc_features.T
plt.matshow(mfcc_features)
plt.title('MFCC')
filterbank_features=filter_features.T
plt.matshow(filter_features)
plt.title('Filter bank')
plt.show()
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27

MFCC效果图如下:

滤波器特征图像如下:

输出结果如下:

MFCC :
 Number of window = 40
每个特征的长度为:  13

 Filter bank: 
 Number of window = 40
每个特征的长度为:  26
1
2
3
4
5
6
7