forked from IDRnD/antispoofing-features
-
Notifications
You must be signed in to change notification settings - Fork 0
/
signal_features.py
284 lines (218 loc) · 9.04 KB
/
signal_features.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
import json
from collections import Counter
from multiprocessing import Pool
from pathlib import Path, PosixPath
import numpy as np
from tqdm import tqdm
from hurst import compute_Hc
from skimage.util import view_as_windows
from librosa.feature import zero_crossing_rate, rms
from config import Config
from tools.feature_extractor import SignalLoader, stat_features
def get_repeats(signal: np.ndarray,
num_features: int = 16,
threshold: int = 1,
dtype=np.uint8):
"""
Counts the number of plateaus of different lengths in the raw signal
Parameters:
signal: np.ndarray - input 1D signal
num_features: int - maximim length of detected repeats
threshold: int - minimum length of detected repeats
dtype: type of extracted features
Returns:
repeat_counter: np.ndarray - plateau features
feature_names: list - names of extracted features
"""
repeat_counter = np.zeros(shape=(num_features,), dtype=dtype)
file_conv = np.convolve(signal, [-1, 1])
file_conv = np.where(file_conv == 0)[0]
file_conv = file_conv - np.arange(start=0, stop=len(file_conv))
counts = np.unique(file_conv, return_counts=True)[1] - threshold
repeats_counter = Counter(counts[(counts > 0) & (counts < num_features + threshold)])
for key, val in repeats_counter.items():
repeat_counter[key - 1] = val
feature_names = [f'repeats_{i}' for i in range(num_features)]
return repeat_counter, feature_names
def get_signal_stats(signal: np.ndarray, signal_features_config: dict):
"""
Extracts various statistics from the raw signal
Parameters:
signal: np.ndarray - input 1D signal
signal_features_config: dict - stat. features that should be extracted
Returns:
features: np.ndarray - extracted stat. features
feature_names: list - names of extracted features
"""
features, feature_names = [], []
file_types = {
'signal': signal,
'abs': np.abs(signal),
'diff': np.diff(signal),
'zero_cross': zero_crossing_rate(signal)[0],
'rms': rms(signal)[0]
}
for signal_feature_name, config in signal_features_config.items():
for stat_feature_key, included in config.items():
if not included:
continue
stat_feature_name = stat_feature_key.split('_')[0]
feature = stat_features[stat_feature_name](file_types[signal_feature_name])
if stat_feature_key == 'mode_val':
feature = feature.mode[0]
elif stat_feature_key == 'mode_cnt':
feature = feature.count[0]
features.append(feature)
name = f'{signal_feature_name}_{stat_feature_key}'
feature_names.append(name)
return features, feature_names
def get_symmetry_diff(signal: np.ndarray, eps: float = 1e-8):
"""
Calculates samples symmetry relative to zero
Parameters:
signal: np.ndarray - input 1D signal
eps: float - epsilon constant
Returns:
features: np.ndarray - extracted stat. features
feature_names: list - names of extracted features
"""
pos_part = signal[signal >= 0]
neg_part = signal[signal < 0]
symmetry_diff = np.abs(pos_part.sum() + neg_part.sum())
min_len = min(len(pos_part), len(neg_part))
pos_part = pos_part[: min_len]
neg_part = neg_part[: min_len]
diff = pos_part + neg_part
num_equal_bins = len(np.where(diff == 0)[0]) / (len(diff) + eps)
num_diff_bins = 1 - num_equal_bins
symmetry_diff = symmetry_diff / (np.abs(signal).max() + eps)
features = [num_equal_bins, num_diff_bins, symmetry_diff]
feature_names = ['symm_num_equal_bins', 'symm_num_diff_bins', 'symm_abs_diff']
return features, feature_names
def get_hurst_exp(signal: np.ndarray):
"""
Extracts Hurst coefficients from the raw signal
Parameters:
signal: np.ndarray - input 1D signal
Returns:
features: np.ndarray - extracted stat. features
feature_names: list - names of extracted features
"""
H, c, _ = compute_Hc(signal, kind='random_walk', simplified=True)
features = np.array([H, c])
feature_names = ['hurst_H', 'hurst_c']
return features, feature_names
def calc_window_stats(chunks, feature, name, num_bins):
if name == 'mode_val':
chunks = feature(chunks, axis=1).mode[0]
elif name == 'mode_cnt':
chunks = feature(chunks, axis=1).count[0]
else:
chunks = feature(chunks, axis=1)
diff = np.abs(np.diff(chunks))
hist = np.histogram(diff, bins=num_bins)
return hist
def get_window_stats(signal: np.ndarray,
win_len: int = 4096,
step: int = 256,
num_bins: int = 5):
"""
Slices the raw signal into a set of windows and calculates statistical parameters for each window
Parameters:
signal: np.ndarray - input 1D signal
win_len: int - length of the window
step: int - step size
num_bins: int - number of bins in the generated histogram
Returns:
features: np.ndarray - extracted stat. features
feature_names: list - names of extracted features
"""
wind_stats_config = {
'min': True,
'max': True,
'mean': True,
'std': True,
'skew': True,
'kurtosis': True,
'mode_val': True,
'mode_count': True,
'iqr': True,
'sem': True,
'mad': True
}
features = []
feature_names = []
chunks = view_as_windows(signal, win_len, step)
for name, included in wind_stats_config.items():
if not included:
continue
feature_name = name.split('_')[0]
feature = stat_features[feature_name]
bin_count, bin_val = calc_window_stats(chunks, feature, name, num_bins)
features.extend([bin_count, bin_val])
names = []
for i in range(1, len(bin_count) + 1):
names.append(f'wind_stats_{name}_cnt_{i}')
for i in range(1, len(bin_val) + 1):
names.append(f'wind_stats_{name}_val_{i}')
feature_names.extend(names)
return np.hstack(features), feature_names
def calc_statistics(paths: list,
feature_name: str,
feature_type: str,
stat_func,
save_path: PosixPath = None,
save_feature_names: bool = False,
dtype=np.float32):
"""
Calculates stat. features on the raw audio signal
Parameters:
paths: np.ndarray - paths to the audio files
feature_name: str - name of the processed feature
feature_type: str - type of the processed feature (train, dev or val)
stat_func: function for extraction of stat. feature
save_path: PosixPath - path for saving of calculated features
save_feature_names: bool - save names of calculated features or not
dtype: type of extracted features
Returns:
statistics: np.ndarray - extracted stat. features
"""
assert feature_type in ('train', 'dev', 'val')
statistics = []
with Pool(processes=Config.num_proc) as p:
with tqdm(total=len(paths)) as pbar:
for stat, names in p.imap(stat_func, paths):
statistics.append(stat)
pbar.update()
if save_path is not None and save_feature_names:
(save_path / 'feature_names').mkdir(exist_ok=True, parents=True)
np.save(save_path / 'feature_names' / f'{feature_name}_names', names)
statistics = np.array(statistics, dtype=dtype)
if save_path is not None:
(save_path / feature_type).mkdir(exist_ok=True, parents=True)
np.save(save_path / f'{feature_type}/{feature_name}', statistics)
return statistics
if __name__ == '__main__':
root_dir = Path(__file__).parent
file_path = root_dir / 'tests' / 'LA_T_1000137.flac'
config_path = root_dir / 'configs'
with open(config_path / 'signal_features.json', 'r') as config:
signal_features_config = json.load(config)
features_config = {
'repeats': {'stat_func': SignalLoader(get_repeats, normalize=True, **{'num_features': 16}), 'dtype': np.uint8},
'stats': {'stat_func': SignalLoader(get_signal_stats, normalize=True, **{'signal_features_config': signal_features_config})},
'symmetry': {'stat_func': SignalLoader(get_symmetry_diff)},
'wind_stats': {'stat_func': SignalLoader(get_window_stats, normalize=True)},
'hurst': {'stat_func': SignalLoader(get_hurst_exp)}
}
paths = [file_path]
feature_type = 'train'
for feature_name, config in features_config.items():
statistics = calc_statistics(paths=paths,
feature_name=feature_name,
feature_type=feature_type,
**config)
statistics_test = np.load(root_dir / 'tests' / 'signal_features' / f'LA_T_1000137_{feature_name}.npy')
assert np.all(statistics == statistics_test), f'Test for {feature_name} not passed'
print(feature_name, statistics.shape)
print('OK')