-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathdata_utils_AASIST_base_train.py
200 lines (152 loc) · 5.93 KB
/
data_utils_AASIST_base_train.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
import numpy as np
import soundfile as sf
import torch
from torch import Tensor
from torch.utils.data import Dataset
import librosa
from random import randrange
import random
import os
___author__ = "Hemlata Tak, Jee-weon Jung, Antani"
__email__ = "tak@eurecom.fr, jeeweon.jung@navercorp.com, antani@eurecom.fr"
def genSpoof_list(dir_meta, is_train=False, is_eval=False):
d_meta = {}
file_list = []
utt_IDs=[]
label_list=[]
with open(dir_meta, "r") as f:
l_meta = f.readlines()
if is_train:
for line in l_meta:
# key,label = line.strip().split(" ")
#_,key =path.split("/")
_, key, _, _, label = line.strip().split() # ugly crap to make it fit to our protocols
file_list.append(key)
d_meta[key] = 1 if label == "bonafide" else 0
return d_meta, file_list
elif is_eval:
for line in l_meta:
# key, label = line.strip().split(" ")
_, key, _, _, label = line.strip().split() # ugly crap to make it fit to our protocols
# _,utt_id =key.split("/")
utt_id = key
utt_IDs.append(utt_id)
file_list.append(key)
label_list.append(label)
d_meta[key] = 1 if label == "bonafide" else 0
return d_meta,file_list,utt_IDs,label_list
else:
for line in l_meta:
# key,label = line.strip().split(" ")
_, key, _, _, label = line.strip().split() # ugly crap to make it fit to our protocols
#_,utt_id =key.split("/")
#utt_ID.append(utt_id)
file_list.append(key)
d_meta[key] = 1 if label == "bonafide" else 0
return d_meta, file_list
def pad(x, max_len=64600):
x_len = x.shape[0]
if x_len >= max_len:
return x[:max_len]
# need to pad
num_repeats = int(max_len / x_len) + 1
padded_x = np.tile(x, (1, num_repeats))[:, :max_len][0]
return padded_x
def pad_random(x: np.ndarray, max_len: int = 64600):
x_len = x.shape[0]
# if duration is already long enough
if x_len >= max_len:
stt = np.random.randint(x_len - max_len)
return x[stt:stt + max_len]
# if too short
num_repeats = int(max_len / x_len) + 1
padded_x = np.tile(x, (num_repeats))[:max_len]
return padded_x
class Dataset_train(Dataset):
def __init__(self,list_IDs, labels, base_dir):
self.list_IDs = list_IDs
self.labels = labels
self.base_dir = base_dir
self.cut = 64600 # take ~4 sec audio (64600 samples)
self.fs = 16000
def __len__(self):
return len(self.list_IDs)
def __getitem__(self, index):
key = self.list_IDs[index]
# X, _ = librosa.load(self.base_dir+key+'.flac', sr=16000)
X, _ = librosa.load(os.path.join(self.base_dir, f'{key}.flac'), sr=16000)
X_pad = pad_random(X, self.cut)
x_inp = Tensor(X_pad)
target = self.labels[key]
return x_inp, target
class Dataset_dev(Dataset):
def __init__(self,list_IDs, labels, base_dir):
self.list_IDs = list_IDs
self.labels = labels
#self.utt_id = UTT_ID
self.base_dir = base_dir
self.cut = 64600 # take ~4 sec audio (64600 samples)
self.fs = 16000
def __len__(self):
return len(self.list_IDs)
def __getitem__(self, index):
key = self.list_IDs[index]
#id_utt=self.utt_id[index]
# X, _ = librosa.load(self.base_dir+key+'.flac', sr=16000)
X, _ = librosa.load(os.path.join(self.base_dir, f'{key}.flac'), sr=16000)
X_pad = pad_random(X, self.cut)
x_inp = Tensor(X_pad)
target = self.labels[key]
return x_inp, target
class Dataset_eval(Dataset):
def __init__(self,list_IDs,labels, UTT_ID, base_dir, att_list):
self.list_IDs = list_IDs
self.utt_id = UTT_ID
self.labels = labels
self.base_dir = base_dir
self.att_list = att_list
self.cut = 64600 # take ~4 sec audio (64600 samples)
self.fs = 16000
def __len__(self):
return len(self.list_IDs)
def __getitem__(self, index):
key = self.list_IDs[index]
id_utt = self.utt_id[index]
att = self.att_list[index]
# X, _ = librosa.load(self.base_dir+key+'.flac', sr=16000) # for debug # wait, you didn't say "dubug" here?
#X, _ = librosa.load(self.base_dir+key, sr=16000)
X, _ = librosa.load(os.path.join(self.base_dir, f'{key}.flac'), sr=16000)
X_pad = pad_random(X, self.cut)
x_inp = Tensor(X_pad)
target = self.labels[key]
return x_inp,target, id_utt,att
class Dataset_ASVspoof2019_devNeval(Dataset):
def __init__(self, list_IDs, base_dir):
"""self.list_IDs : list of strings (each string: utt key),
"""
self.list_IDs = list_IDs
self.base_dir = base_dir
self.cut = 64600 # take ~4 sec audio (64600 samples)
def __len__(self):
return len(self.list_IDs)
def __getitem__(self, index):
key = self.list_IDs[index]
X, _ = librosa.load(self.base_dir+'flac/'+key+'.flac', sr=16000)
X_pad = pad(X, self.cut)
x_inp = Tensor(X_pad)
return x_inp, key
class Dataset_ASVspoof2021_eval(Dataset):
def __init__(self, list_IDs, base_dir):
'''self.list_IDs : list of strings (each string: utt key),
'''
self.list_IDs = list_IDs
self.base_dir = base_dir
def __len__(self):
return len(self.list_IDs)
def __getitem__(self, index):
self.cut=64600 # take ~4 sec audio (64600 samples)
key = self.list_IDs[index]
X, fs = librosa.load(self.base_dir+'flac/'+key+'.flac', sr=16000)
X_pad = pad(X,self.cut)
x_inp = Tensor(X_pad)
return x_inp,key