-
Notifications
You must be signed in to change notification settings - Fork 8
/
generate_samples.py
339 lines (273 loc) · 10.4 KB
/
generate_samples.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
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
# Routines to select groups of speakers from the dataset
# Copyright 2020 Robin Scheibler
#
# Permission is hereby granted, free of charge, to any person obtaining a copy
# of this software and associated documentation files (the "Software"), to deal
# in the Software without restriction, including without limitation the rights
# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
# copies of the Software, and to permit persons to whom the Software is
# furnished to do so, subject to the following conditions:
#
# The above copyright notice and this permission notice shall be included in all
# copies or substantial portions of the Software.
#
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
# SOFTWARE.
"""
This script will create samples composed of concatenations of utterances
from a same speaker
"""
import json
import os
import pickle
import random
from itertools import combinations, permutations
import numpy as np
from scipy.io import wavfile
import pyroomacoustics as pra
def sampling(
num_subsets, num_speakers, metadata_file, gender_balanced=False, seed=None
):
"""
This function will pick automatically and at random subsets
speech samples from a list generated using this file.
Parameters
----------
num_subsets: int
Number of subsets to create
num_speakers: int
Number of distinct speakers desired in a subset
metadata_file: str
Location of the metadata file
gender_balanced: bool, optional
If True, the subsets will have a the same number of male/female speakers
when `num_speakers` is even, and one extra male, when `num_speakers` is odd.
Default is `False`.
seed: int, optional
When a seed is provided, the random number generator is fixed to a deterministic
state. This is useful for getting consistently the same set of speakers.
The initial state of the random number generator is restored at the end of the function.
When not provided, the random number generator is used without setting the seed.
Returns
-------
A list of `num_subsets` lists of wav filenames, each containing `num_speakers` entries.
"""
# save current numpy RNG state and set a known seed
if seed is not None:
rng_state = random.getstate()
random.seed(seed)
samples_dir = os.path.split(metadata_file)[0]
# load metadata
with open(metadata_file, "r") as f:
metadata = json.load(f)
if gender_balanced:
samples = [metadata["sorted"]["male"], metadata["sorted"]["female"]]
numbers = [num_speakers // 2, num_speakers - num_speakers // 2]
else:
samples = [metadata["sorted"]["male"].copy()]
samples[0].update(metadata["sorted"]["female"])
numbers = [num_speakers]
# now create a list of list of files
all_lists = [[] for i in range(num_subsets)]
for S, num in zip(samples, numbers):
speakers = list(S.keys())
random.shuffle(speakers)
all_combs = list(combinations(speakers, num))
for sub_list in all_lists:
spkrs = list(random.choice(all_combs))
random.shuffle(spkrs)
for spkr in spkrs:
sub_list.append(os.path.join(samples_dir, random.choice(S[spkr])))
# restore numpy RNG former state
if seed is not None:
random.setstate(rng_state)
return all_lists
def wav_read_center(wav_list, center=True, seed=None):
"""
Read a bunch of wav files, equalize their length
and puts them in a numpy array
Parameters
----------
wav_list: list of str
A list of file names, the file names should be of format wav and monaural
center: bool, optional
When True (default), the signals will be centered, otherwise, only their end will be zero padded
seed: int
Provides a seed for the random number generator. When this is provided,
center option is ignored and the beginning of segments is placed at
random within the maximum length available
Returns
-------
ndarray (n_files, n_samples)
A 2D array that contains one signal per row
"""
if seed is not None:
rng_state = np.random.get_state()
np.random.seed(seed)
rows = []
fs = None
for fn in wav_list:
fs_loc, data = wavfile.read(fn)
data = data.astype(np.float)
if fs is None:
fs = fs_loc
if fs != fs_loc:
raise ValueError("All the files should have the same sampling frequency")
if data.ndim > 1:
import warnings
warnings.warn("Discarding extra channels of non-monaural file")
data = data[:, 0]
rows.append(data)
max_len = np.max([d.shape[0] for d in rows])
output = np.zeros((len(rows), max_len), dtype=rows[0].dtype)
for r, row in enumerate(rows):
if seed is not None:
slack = max_len - row.shape[0]
if slack > 0:
b = np.random.randint(0, max_len - row.shape[0])
else:
b = 0
elif center:
b = (max_len - row.shape[0]) // 2
else:
b = 0
output[r, b : b + row.shape[0]] = row
if seed is not None:
np.random.set_state(rng_state)
output /= 2 ** 15
return output
def generate_samples(n_speakers_per_sex, n_samples, duration, seed, cmudir, output_dir):
"""
Generate the dataset from the CMU Arctic Corpus
Parameters
----------
n_speakers_per_sex: int
The number of speakers of each sex to include
n_samples: int
The number of samples to generate per speaker
duration: float
The minimum duration of one sample
seed: int
The seed for the random number generator
cmudir: str
The location of the CMU Arctic Corpus
output_dir: str
The location where to save the new dataset
"""
random.seed(seed)
filename = "cmu_arctic_{sex}_{spkr}_{ind}.wav"
if not os.path.exists(output_dir):
os.mkdir(output_dir)
# load database and cache locally, or load cached db if existing
cmu_cache_file = "cmu_arctic.dat"
if not os.path.exists(cmu_cache_file):
if cmudir is None:
print("Cache DB doesn" "t exist, downloading. This could take long...")
cmu_arctic = pra.datasets.CMUArcticCorpus(download=True)
else:
print("Cache DB doesn" "t exist, loading from scratch")
cmu_arctic = pra.datasets.CMUArcticCorpus(basedir=cmudir)
with open(cmu_cache_file, "wb") as f:
pickle.dump(cmu_arctic, f)
else:
print("Cache DB exists, loading")
with open(cmu_cache_file, "rb") as f:
cmu_arctic = pickle.load(f)
# get data type and sampling frequency of dataset
dtype = cmu_arctic.sentences[0].data.dtype
fs = cmu_arctic.sentences[0].fs
# a blank segment to insert between sentences
lmin, lmax = int(0.5) * fs, int(2.5) * fs
def new_blank():
return np.zeros(lmin + random.randint(lmin, lmax), dtype=dtype)
# keep track of metadata in a file
metadata = {
"generation_args": {
"n_speakers_per_sex": n_speakers_per_sex,
"n_samples": n_samples,
"duration": duration,
"seed": seed,
},
"fs": fs,
"files": [],
"transcripts": {},
"sorted": {},
}
# iterate over different speakers to create the concatenated sentences
for sex in ["male", "female"]:
ds_sex = cmu_arctic.filter(sex=sex)
speakers = list(ds_sex.info["speaker"].keys())
metadata["sorted"][sex] = {}
for speaker in speakers[:n_speakers_per_sex]:
ds_spkr = ds_sex.filter(speaker=speaker)
random.shuffle(ds_spkr.sentences)
sentence_iter = iter(ds_spkr.sentences)
metadata["sorted"][sex][speaker] = []
n = 0
while n < n_samples:
new_sentence = [new_blank()]
L = len(new_sentence[-1])
texts = []
while L < duration * fs:
# add new sample audio
s = next(sentence_iter)
new_sentence.append(s.data)
L += s.data.shape[0]
# add the text to transcript
texts.append(s.meta.text)
new_sentence.append(new_blank())
L += len(new_sentence[-1])
fn = filename.format(sex=sex, spkr=speaker, ind=n + 1)
wavfile.write(
os.path.join(output_dir, fn), fs, np.concatenate(new_sentence),
)
metadata["sorted"][sex][speaker].append(fn)
metadata["files"].append(fn)
metadata["transcripts"][fn] = " ".join(texts)
n += 1
with open(os.path.join(output_dir, "metadata.json"), "w") as f:
json.dump(metadata, f, indent=4)
if __name__ == "__main__":
import argparse
parser = argparse.ArgumentParser(
description=("Generates the dataset from the CMU ARCTIC speech corpus")
)
parser.add_argument(
"-s", "--n_speakers", type=int, default=7, help="Number of speakers per sex"
)
parser.add_argument(
"-n", "--n_samples", type=int, default=10, help="Number of samples per speaker"
)
parser.add_argument(
"-d", "--duration", type=float, default=15, help="Minimum duration requested"
)
parser.add_argument(
"--seed",
type=int,
default=38448,
help="The seed for the random number generator",
)
parser.add_argument(
"-c", "--cmudir", type=str, help="Directory of CMU ARCTIC corpus, if available"
)
parser.add_argument(
"-o",
"--output",
type=str,
default=".",
help="Output directory where to save all the files",
)
args = parser.parse_args()
generate_samples(
args.n_speakers,
args.n_samples,
args.duration,
args.seed,
args.cmudir,
args.output,
)