-
Notifications
You must be signed in to change notification settings - Fork 1.4k
/
Copy pathprepare_data.py
248 lines (219 loc) · 7.61 KB
/
prepare_data.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
"""
The .csv preparation functions for WSJ0-Mix.
Author
* Cem Subakan 2020
"""
import csv
import os
def prepare_wsjmix(
datapath,
savepath,
n_spks=2,
skip_prep=False,
librimix_addnoise=False,
fs=8000,
):
"""
Prepared wsj2mix if n_spks=2 and wsj3mix if n_spks=3.
Arguments:
----------
datapath (str) : path for the wsj0-mix dataset.
savepath (str) : path where we save the csv file.
n_spks (int): number of speakers
skip_prep (bool): If True, skip data preparation
librimix_addnoise: If True, add whamnoise to librimix datasets
"""
if skip_prep:
return
if "wsj" in datapath:
if n_spks == 2:
assert (
"2speakers" in datapath
), "Inconsistent number of speakers and datapath"
create_wsj_csv(datapath, savepath)
elif n_spks == 3:
assert (
"3speakers" in datapath
), "Inconsistent number of speakers and datapath"
create_wsj_csv_3spks(datapath, savepath)
else:
raise ValueError("Unsupported Number of Speakers")
else:
print("Creating a csv file for a custom dataset")
create_custom_dataset(datapath, savepath)
def create_custom_dataset(
datapath,
savepath,
dataset_name="custom",
set_types=["train", "valid", "test"],
folder_names={
"source1": "source1",
"source2": "source2",
"mixture": "mixture",
},
):
"""
This function creates the csv file for a custom source separation dataset
"""
for set_type in set_types:
mix_path = os.path.join(datapath, set_type, folder_names["mixture"])
s1_path = os.path.join(datapath, set_type, folder_names["source1"])
s2_path = os.path.join(datapath, set_type, folder_names["source2"])
files = os.listdir(mix_path)
mix_fl_paths = [os.path.join(mix_path, fl) for fl in files]
s1_fl_paths = [os.path.join(s1_path, fl) for fl in files]
s2_fl_paths = [os.path.join(s2_path, fl) for fl in files]
csv_columns = [
"ID",
"duration",
"mix_wav",
"mix_wav_format",
"mix_wav_opts",
"s1_wav",
"s1_wav_format",
"s1_wav_opts",
"s2_wav",
"s2_wav_format",
"s2_wav_opts",
"noise_wav",
"noise_wav_format",
"noise_wav_opts",
]
with open(
os.path.join(savepath, dataset_name + "_" + set_type + ".csv"),
"w",
encoding="utf-8",
) as csvfile:
writer = csv.DictWriter(csvfile, fieldnames=csv_columns)
writer.writeheader()
for i, (mix_path, s1_path, s2_path) in enumerate(
zip(mix_fl_paths, s1_fl_paths, s2_fl_paths)
):
row = {
"ID": i,
"duration": 1.0,
"mix_wav": mix_path,
"mix_wav_format": "wav",
"mix_wav_opts": None,
"s1_wav": s1_path,
"s1_wav_format": "wav",
"s1_wav_opts": None,
"s2_wav": s2_path,
"s2_wav_format": "wav",
"s2_wav_opts": None,
}
writer.writerow(row)
def create_wsj_csv(datapath, savepath):
"""
This function creates the csv files to get the speechbrain data loaders for the wsj0-2mix dataset.
Arguments:
datapath (str) : path for the wsj0-mix dataset.
savepath (str) : path where we save the csv file
"""
for set_type in ["tr", "cv", "tt"]:
mix_path = os.path.join(datapath, "wav8k/min/" + set_type + "/mix/")
s1_path = os.path.join(datapath, "wav8k/min/" + set_type + "/s1/")
s2_path = os.path.join(datapath, "wav8k/min/" + set_type + "/s2/")
files = os.listdir(mix_path)
mix_fl_paths = [mix_path + fl for fl in files]
s1_fl_paths = [s1_path + fl for fl in files]
s2_fl_paths = [s2_path + fl for fl in files]
csv_columns = [
"ID",
"duration",
"mix_wav",
"mix_wav_format",
"mix_wav_opts",
"s1_wav",
"s1_wav_format",
"s1_wav_opts",
"s2_wav",
"s2_wav_format",
"s2_wav_opts",
]
with open(
savepath + "/wsj_" + set_type + ".csv",
"w",
newline="",
encoding="utf-8",
) as csvfile:
writer = csv.DictWriter(csvfile, fieldnames=csv_columns)
writer.writeheader()
for i, (mix_path, s1_path, s2_path) in enumerate(
zip(mix_fl_paths, s1_fl_paths, s2_fl_paths)
):
row = {
"ID": i,
"duration": 1.0,
"mix_wav": mix_path,
"mix_wav_format": "wav",
"mix_wav_opts": None,
"s1_wav": s1_path,
"s1_wav_format": "wav",
"s1_wav_opts": None,
"s2_wav": s2_path,
"s2_wav_format": "wav",
"s2_wav_opts": None,
}
writer.writerow(row)
def create_wsj_csv_3spks(datapath, savepath):
"""
This function creates the csv files to get the speechbrain data loaders for the wsj0-3mix dataset.
Arguments:
datapath (str) : path for the wsj0-mix dataset.
savepath (str) : path where we save the csv file
"""
for set_type in ["tr", "cv", "tt"]:
mix_path = os.path.join(datapath, "wav8k/min/" + set_type + "/mix/")
s1_path = os.path.join(datapath, "wav8k/min/" + set_type + "/s1/")
s2_path = os.path.join(datapath, "wav8k/min/" + set_type + "/s2/")
s3_path = os.path.join(datapath, "wav8k/min/" + set_type + "/s3/")
files = os.listdir(mix_path)
mix_fl_paths = [mix_path + fl for fl in files]
s1_fl_paths = [s1_path + fl for fl in files]
s2_fl_paths = [s2_path + fl for fl in files]
s3_fl_paths = [s3_path + fl for fl in files]
csv_columns = [
"ID",
"duration",
"mix_wav",
"mix_wav_format",
"mix_wav_opts",
"s1_wav",
"s1_wav_format",
"s1_wav_opts",
"s2_wav",
"s2_wav_format",
"s2_wav_opts",
"s3_wav",
"s3_wav_format",
"s3_wav_opts",
]
with open(
savepath + "/wsj_" + set_type + ".csv",
"w",
newline="",
encoding="utf-8",
) as csvfile:
writer = csv.DictWriter(csvfile, fieldnames=csv_columns)
writer.writeheader()
for i, (mix_path, s1_path, s2_path, s3_path) in enumerate(
zip(mix_fl_paths, s1_fl_paths, s2_fl_paths, s3_fl_paths)
):
row = {
"ID": i,
"duration": 1.0,
"mix_wav": mix_path,
"mix_wav_format": "wav",
"mix_wav_opts": None,
"s1_wav": s1_path,
"s1_wav_format": "wav",
"s1_wav_opts": None,
"s2_wav": s2_path,
"s2_wav_format": "wav",
"s2_wav_opts": None,
"s3_wav": s3_path,
"s3_wav_format": "wav",
"s3_wav_opts": None,
}
writer.writerow(row)