-
Notifications
You must be signed in to change notification settings - Fork 13
/
data_split.py
254 lines (217 loc) · 10.9 KB
/
data_split.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
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
import os
import random
import shutil
root = '/ceph/datasets/VCTK-Corpus/wav48/'
# root = '/ceph/home/yangsc21/Python/autovc/autovc/wavs/'
my_root = '/ceph/home/yangsc21/Python/VCTK/'
training_log = my_root + 'split_training_log.txt'
testing_log = my_root + 'split_testing_log.txt'
def get_filelist(dir):
with open(training_log, 'w', encoding='utf-8') as f_train:
with open(testing_log, 'w', encoding='utf-8') as f_test:
for name in os.listdir(dir):
f_train.write(name + '\n')
f_test.write(name + '\n')
wav_list = os.listdir(os.path.join(dir, name))
random.shuffle(wav_list)
split_point = int(len(wav_list) * 0.9)
for wav in wav_list[:split_point]:
f_train.write(wav + '\n')
for wav in wav_list[split_point:]:
f_test.write(wav + '\n')
def copy_train_file():
with open(training_log, 'r', encoding='utf-8') as f_train:
for line in f_train.readlines():
if line.strip() != '' and line.strip()[-4:] != '.wav':
if not os.path.exists(os.path.join(my_root, 'training_data', line.strip())):
os.makedirs(os.path.join(my_root, 'training_data', line.strip()))
if line.strip()[-4:] == '.wav':
shutil.copy(os.path.join(root, line[:4], line.strip()), os.path.join(my_root, 'training_data', line[:4]))
def copy_test_file():
with open(testing_log, 'r', encoding='utf-8') as f_test:
for line in f_test.readlines():
if line.strip() != '' and line.strip()[-4:] != '.wav':
if not os.path.exists(os.path.join(my_root, 'testing_data', line.strip())):
os.makedirs(os.path.join(my_root, 'testing_data', line.strip()))
if line.strip()[-4:] == '.wav':
shutil.copy(os.path.join(root, line[:4], line.strip()), os.path.join(my_root, 'testing_data', line[:4]))
if __name__ == '__main__':
print('get filelist...')
get_filelist(root)
print('copy train file...')
copy_train_file()
print('copy test file...')
copy_test_file()
import os
import random
import shutil
import numpy as np
VCTK_txt = '/ceph/datasets/VCTK-Corpus/txt/'
log_speaker_path = '/ceph/home/yangsc21/Python/VCTK/wav16/split_speaker.txt'
# log_dim_path = '/ceph/home/yangsc21/Python/VCTK/mel_dim.txt'
log_dim_path = '/ceph/home/yangsc21/Python/VCTK/wav16/mel_dim.txt'
log_dim_9_path = '/ceph/home/yangsc21/Python/VCTK/wav16/mel_dim_6.txt'
log_dim_100_path = '/ceph/home/yangsc21/Python/VCTK/wav16/mel_dim_100.txt'
log_test_path = '/ceph/home/yangsc21/Python/VCTK/wav16/select_from_test.txt'
log_train_path = '/ceph/home/yangsc21/Python/VCTK/wav16/select_from_train.txt'
def split_speaker(root, num_train):
speaker_list = os.listdir(root)
random.shuffle(speaker_list)
print(len(speaker_list))
with open(log_speaker_path, 'w', encoding='utf-8') as f:
f.write('train:\n')
for i in speaker_list[:num_train]:
f.write(i + ' ')
f.write('\n')
f.write('test:\n')
for i in speaker_list[num_train:]:
f.write(i + ' ')
def mel_dim(root):
with open(log_speaker_path, 'r', encoding='utf-8') as f:
speaker_list = f.read().strip().split('\n')
train_speaker_list = speaker_list[1].strip().split(' ')
test_speaker_list = speaker_list[3].strip().split(' ')
# print(train_speaker_list, test_speaker_list)
with open(log_dim_path, 'w', encoding='utf-8') as ff:
ff.write('train:\n')
i = 0
for speak in train_speaker_list:
print(str(i) + '\t' + speak)
for mel in os.listdir(os.path.join(root, speak)):
ff.write(str(i) + '\t')
ff.write(mel + '\t')
ff.write(str(np.load(os.path.join(root, speak, mel)).shape[0]))
ff.write('\n')
i += 1
i = 0
ff.write('test:\n')
for speak in test_speaker_list:
print(str(i) + '\t' + speak)
for mel in os.listdir(os.path.join(root, speak)):
ff.write(str(i) + '\t')
ff.write(mel + '\t')
ff.write(str(np.load(os.path.join(root, speak, mel)).shape[0]))
ff.write('\n')
i += 1
def select_from_test(): # select_same_size
speakers = [[] for _ in range(6)]
with open(log_test_path, 'w', encoding='utf-8') as out:
with open(log_dim_9_path, 'r', encoding='utf-8') as f:
for line in f.readlines():
if line.strip() == 'test:':
continue
(id, rideo, dim) = line.strip().split('\t')
speakers[int(id)].append([rideo, int(dim)])
for speaker_i in range(len(speakers) - 1):
print(speaker_i)
for speaker_j in range(speaker_i + 1, len(speakers)):
for item_i in speakers[speaker_i]:
for item_j in speakers[speaker_j]:
if item_i[0][5:8] == item_j[0][5:8] and item_i[1] == item_j[1]:
with open(os.path.join(VCTK_txt, item_i[0][:4], item_i[0][:8] + '.txt'), 'r', encoding='utf-8') as f_i:
with open(os.path.join(VCTK_txt, item_j[0][:4], item_j[0][:8] + '.txt'), 'r', encoding='utf-8') as f_j:
if f_i.read().strip() == f_j.read().strip():
# print(item_i, item_j)
out.write(item_i[0] + '\t' + item_j[0] + '\t' + str(item_i[1]) + '\n')
def select_from_train(): # select_same_size
speakers = [[] for _ in range(100)]
with open(log_train_path, 'w', encoding='utf-8') as out:
with open(log_dim_100_path, 'r', encoding='utf-8') as f:
for line in f.readlines():
if line.strip() == 'train:':
continue
(id, rideo, dim) = line.strip().split('\t')
speakers[int(id)].append([rideo, int(dim)])
for speaker_i in range(len(speakers) - 1):
print(speaker_i)
for speaker_j in range(speaker_i + 1, len(speakers)):
for item_i in speakers[speaker_i]:
for item_j in speakers[speaker_j]:
if item_i[0][:4] == 'p315' or item_j[0][:4] == 'p315':
continue
if item_i[0][5:8] == item_j[0][5:8] and item_i[1] == item_j[1]:
with open(os.path.join(VCTK_txt, item_i[0][:4], item_i[0][:8] + '.txt'), 'r', encoding='utf-8') as f_i:
with open(os.path.join(VCTK_txt, item_j[0][:4], item_j[0][:8] + '.txt'), 'r', encoding='utf-8') as f_j:
if f_i.read().strip() == f_j.read().strip():
# print(i, item_i, item_j)
out.write(item_i[0] + '\t' + item_j[0] + '\t' + str(item_i[1]) + '\n')
def movefile(srcfile, dstfile):
if not os.path.isfile(srcfile):
print(srcfile + " not exist!")
else:
fpath, fname = os.path.split(dstfile) # 分离文件名和路径
if not os.path.exists(fpath):
os.makedirs(fpath) # 创建路径
shutil.move(srcfile, dstfile) # 移动文件
def move_rideo():
source_path = '/ceph/home/yangsc21/Python/VCTK/wav16/spmel_6/'
target_path = '/ceph/home/yangsc21/Python/VCTK/wav16/spmel_6_/'
with open(log_test_path, 'r', encoding='utf-8') as f:
for line in f.readlines():
(source1, source2, dim) = line.strip().split('\t')
try:
movefile(os.path.join(source_path, source1[:4], source1), os.path.join(target_path, source1[:4], source1))
movefile(os.path.join(source_path, source2[:4], source2), os.path.join(target_path, source2[:4], source2))
except:
continue
source_path = '/ceph/home/yangsc21/Python/VCTK/wav16/spmel_100/'
target_path = '/ceph/home/yangsc21/Python/VCTK/wav16/spmel_100_/'
with open(log_train_path, 'r', encoding='utf-8') as f:
for line in f.readlines():
(source1, source2, dim) = line.strip().split('\t')
try:
movefile(os.path.join(source_path, source1[:4], source1),
os.path.join(target_path, source1[:4], source1))
movefile(os.path.join(source_path, source2[:4], source2),
os.path.join(target_path, source2[:4], source2))
except:
continue
def crop_len():
root = "/ceph/home/yangsc21/Python/spmel_100_crop/"
log = "/ceph/home/yangsc21/Python/VCTK/wav16/mel_dim_100.txt"
out_log = "/ceph/home/yangsc21/Python/mel_dim_100_crop.txt"
with open(log, 'r', encoding='utf-8') as f1:
with open(out_log, 'w', encoding='utf-8') as f2:
for line in f1.readlines():
if line.strip() == 'train:':
f2.write(line)
continue
(id, rideo, dim) = line.strip().split('\t')
if int(dim) <= 128 * 3:
f2.write(line)
else:
print(line.strip())
try:
os.remove(os.path.join(root, rideo[:4], rideo))
except:
continue
def crop_len_test():
root = "/ceph/home/yangsc21/Python/spmel_6_crop/"
log = "/ceph/home/yangsc21/Python/VCTK/wav16/mel_dim_6.txt"
out_log = "/ceph/home/yangsc21/Python/mel_dim_6_crop.txt"
with open(log, 'r', encoding='utf-8') as f1:
with open(out_log, 'w', encoding='utf-8') as f2:
for line in f1.readlines():
if line.strip() == 'test:':
f2.write(line)
continue
(id, rideo, dim) = line.strip().split('\t')
if int(dim) <= 128 * 3:
f2.write(line)
else:
print(line.strip())
try:
os.remove(os.path.join(root, rideo[:4], rideo))
except:
continue
if __name__ == '__main__':
# root = '/ceph/home/yangsc21/Python/autovc/autovc/test_spmel/'
# root = '/ceph/home/yangsc21/Python/VCTK/wav48_mel/'
# root = "/ceph/home/yangsc21/Python/VCTK/wav16/spmel/"
# split_speaker(root, 100) # 100 for training, total 109 speakers
# mel_dim(root)
# select_from_test()
# select_from_train()
# move_rideo()
# crop_len()
crop_len_test()