某些音频是双方对话,有可能需要对音频作通道的分离。
示例代码如下:
1 #!/usr/bin/env python
2 # -*- coding: utf-8 -*-
3 """
4 音频双通道分离
5 """
6 import sys
7 import numpy as np
8 from scipy.io import wavfile
9 from converter import mp3_to_wav
10
11
12 def split_channel(wav_path, left_wav_path, right_wav_path):
13 """
14 通道分离
15 :param wav_path: wav音频的路径
16 :param left_wav_path: 左声道的wav音频路径
17 :param right_wav_path: 右声道的wav音频路径
18 :return None:
19 """
20 try:
21 sample_rate, wav_data = wavfile.read(wav_path)
22 left = []
23 right = []
24 for item in wav_data:
25 left.append(item[0])
26 right.append(item[1])
27 wavfile.write(left_wav_path, sample_rate, np.array(left))
28 wavfile.write(right_wav_path, sample_rate, np.array(right))
29 except IOError as e:
30 print('error is %s' % str(e))
31 except:
32 print('other error', sys.exc_info())
33
34
35 if __name__ == '__main__':
36 mp3_to_wav('input/test.mp3', 'tmp/tmp.wav')
37 split_channel('tmp/tmp.wav', 'output/left.wav', 'output/right.wav')
其中调用了一个自定义的库,converter.py 加入了代码:
1 from pydub import AudioSegment
2
3
4 def mp3_to_wav(source, destin):
5 """
6 mp3 转 wav
7 :param source:
8 :param destin:
9 :return None:
10 """
11 data = AudioSegment.from_mp3(source)
12 data.export(destin, format='wav')
本代码示例可以生成两个通道分离后的 wav 文件。
注意:如果源文件的格式是 wav 文件,可以省掉转换格式的过程。