-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathwav_aligner.py
115 lines (104 loc) · 3.63 KB
/
wav_aligner.py
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
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
"""
该程序将录同一人同一段语音的两端录音对齐
输入：两个不同长度wav文件 001.wav 002.wav
输出：两个相同长度wav文件 001.wav 002.wav
"""
from scipy.io import wavfile
import scipy.signal as signal
import numpy as np
import os
"""
判断声音真正开始的地方,判断声音结束的地方
通过大于前两帧两倍的参数开始
"""
def _get_path(path):
file_path=[]
for root, dirs, files in os.walk(path):
files.sort()
for file in files:
if file.endswith(".wav"):
file_path.append(os.path.join(root,file))
# file_path.append(file)
return file_path
def _wavcut(wavpath):
rate, data = wavfile.read(wavpath)
if rate != 8000:
result=int(data.shape[0]/rate*8000)
if result%2!=0:#偶数算的快
result+=1
data=signal.resample(data,result)
rate = 8000
#取前1600采样点也就是0.2s的平均值
headframe=np.abs(data[0:1600])
headthresh=np.mean(headframe)
tailframe=np.abs(data[-1600:])
tailthresh = np.mean(tailframe)
end=len(data)
for i in range(int(len(data)/10)):
if np.mean(np.abs(data[i:i+80]))>headthresh*10:
begin=i
break
for i in range(end,int(0.9*end),-1):
if np.mean(np.abs(data[i-80:i]))>tailthresh*10:
end=i
break
newdata=data[begin:end]
return newdata,rate
def _aliggner(data1,data2):
len1=len(data1)
len2=len(data2)
difflen = len1 - len2
if difflen==0:
return data1, data2
elif abs(difflen)>1000:
begin=0
if difflen>0:
difflen=len1-len2
maxsum=np.sum(np.abs(data1[0:(len1-difflen)]-data2))
for i in range(0,difflen,int(difflen/100)):
datasum=np.sum(np.abs(data1[i:(i+len2)]-data2))
if datasum>maxsum:
maxsum=datasum
begin=i
data1=data1[begin:begin+len2]
else:
difflen = len2 - len1
maxsum = np.sum(np.abs(data2[0:(len2 - difflen)] - data1))
for i in range(0, difflen, int(difflen / 100)):
datasum = np.sum(np.abs(data2[i:(i + len1)] - data1))
if datasum > maxsum:
maxsum = datasum
begin = i
data2 = data1[begin:begin + len1]
else:
if difflen>0:
data1=data1[0:len2]
else:
data2=data2[0:len1]
return data1, data2
def main():
# channelpath1="E:\\Data\\phone"
# channelpath2 = "E:\\Data\\app"
# newfiledir="C:\\Users\\liuxk\\Desktop\\实验数据\\fftconvert\\wav\\"
channelpath1 = "E:\\Data\\phone"
channelpath2 = "E:\\Data\\app"
newfiledir = "C:\\Users\\liuxk\\Desktop\\实验数据\\fftconvert\\wav\\"
file_list1=_get_path(channelpath1)
file_list2 = _get_path(channelpath2)
if len(file_list1)!=len(file_list2):
raise Exception("文件数目不对应",)
for i in range(51,len(file_list1),1):
wavpath1=file_list1[i]
wavpath2=file_list2[i]
data1,rate1 = _wavcut(wavpath1)
data2,rate2 = _wavcut(wavpath2)
data1, data2=_aliggner(data1, data2)
data1 = data1.astype(np.int16)
data2 = data2.astype(np.int16)
newwavpath1=newfiledir+wavpath1.split("\\")[-1].replace(".wav","_1new.wav")
newwavpath2=newfiledir+wavpath2.split("\\")[-1].replace(".wav","_2new.wav")
wavfile.write(newwavpath1, rate1, data1)
wavfile.write(newwavpath2, rate2, data2)
print("completed: {com}%".format(com=i/len(file_list1)*100))
if __name__ == '__main__':
main()