-
Notifications
You must be signed in to change notification settings - Fork 2
/
data_utils.py
185 lines (149 loc) · 7.07 KB
/
data_utils.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
import os
import random
import numpy as np
import torch
import torch.utils.data
import utils
from modules.mel_processing import mel_spectrogram_torch
from utils import load_filepaths_and_text, load_wav_to_torch
# import h5py
"""Multi speaker version"""
class TextAudioSpeakerLoader(torch.utils.data.Dataset):
"""
1) loads audio, speaker_id, text pairs
2) normalizes text and converts them to sequences of integers
3) computes spectrograms from audio files.
"""
def __init__(self, audiopaths, hparams, all_in_mem: bool = False, vol_aug: bool = True):
self.audiopaths = load_filepaths_and_text(audiopaths)
self.hparams = hparams
self.max_wav_value = hparams.data.max_wav_value
self.sampling_rate = hparams.data.sampling_rate
self.filter_length = hparams.data.filter_length
self.hop_length = hparams.data.hop_length
self.win_length = hparams.data.win_length
self.unit_interpolate_mode = hparams.data.unit_interpolate_mode
self.sampling_rate = hparams.data.sampling_rate
self.use_sr = hparams.train.use_sr
self.spec_len = hparams.train.max_speclen
self.spk_map = hparams.spk
self.vol_emb = hparams.model.vol_embedding
self.num_mels = hparams.data.n_mel_channels
self.mel_fmin = hparams.data.mel_fmin
self.mel_fmax = hparams.data.mel_fmax
self.vol_aug = hparams.train.vol_aug and vol_aug
random.seed(1234)
random.shuffle(self.audiopaths)
self.all_in_mem = all_in_mem
if self.all_in_mem:
self.cache = [self.get_audio(p[0]) for p in self.audiopaths]
def get_audio(self, filename):
filename = filename.replace("\\", "/")
audio, sampling_rate = load_wav_to_torch(filename)
if sampling_rate != self.sampling_rate:
raise ValueError(
"Sample Rate not match. Expect {} but got {} from {}".format(
self.sampling_rate, sampling_rate, filename))
audio_norm = audio / self.max_wav_value
audio_norm = audio_norm.unsqueeze(0)
spec_filename = filename.replace(".wav", ".mel.pt")
# Ideally, all data generated after Mar 25 should have .mel.pt
if os.path.exists(spec_filename):
spec = torch.load(spec_filename)
else:
spec = mel_spectrogram_torch(audio_norm, self.filter_length,self.num_mels,
self.sampling_rate, self.hop_length, self.win_length,self.mel_fmin,self.mel_fmax,
center=False)
spec = torch.squeeze(spec, 0)
torch.save(spec, spec_filename)
spk = filename.split("/")[-2]
spk = torch.LongTensor([self.spk_map[spk]])
f0, uv = np.load(filename + ".f0.npy",allow_pickle=True)
f0 = torch.FloatTensor(np.array(f0,dtype=float))
uv = torch.FloatTensor(np.array(uv,dtype=float))
c = torch.load(filename+ ".soft.pt")
c = utils.repeat_expand_2d(c.squeeze(0), f0.shape[0], mode=self.unit_interpolate_mode)
if self.vol_emb:
volume_path = filename + ".vol.npy"
volume = np.load(volume_path)
volume = torch.from_numpy(volume).float()
else:
volume = None
lmin = min(c.size(-1), spec.size(-1))
assert abs(c.size(-1) - spec.size(-1)) < 3, (c.size(-1), spec.size(-1), f0.shape, filename)
assert abs(audio_norm.shape[1]-lmin * self.hop_length) < 3 * self.hop_length
spec, c, f0, uv = spec[:, :lmin], c[:, :lmin], f0[:lmin], uv[:lmin]
audio_norm = audio_norm[:, :lmin * self.hop_length]
if volume is not None:
volume = volume[:lmin]
return c, f0, spec, audio_norm, spk, uv, volume
def random_slice(self, c, f0, spec, audio_norm, spk, uv, volume):
# if spec.shape[1] < 30:
# print("skip too short audio:", filename)
# return None
if random.choice([True, False]) and self.vol_aug and volume is not None:
max_amp = float(torch.max(torch.abs(audio_norm))) + 1e-5
max_shift = min(1, np.log10(1/max_amp))
log10_vol_shift = random.uniform(-1, max_shift)
audio_norm = audio_norm * (10 ** log10_vol_shift)
volume = volume * (10 ** log10_vol_shift)
spec = mel_spectrogram_torch(audio_norm, self.filter_length,self.num_mels,
self.sampling_rate, self.hop_length, self.win_length,self.mel_fmin,self.mel_fmax,
center=False)[0]
if spec.shape[1] > 800:
start = random.randint(0, spec.shape[1]-800)
end = start + 790
spec, c, f0, uv = spec[:, start:end], c[:, start:end], f0[start:end], uv[start:end]
audio_norm = audio_norm[:, start * self.hop_length : end * self.hop_length]
if volume is not None:
volume = volume[start:end]
return c, f0, spec, audio_norm, spk, uv,volume
def __getitem__(self, index):
if self.all_in_mem:
return self.random_slice(*self.cache[index])
else:
return self.random_slice(*self.get_audio(self.audiopaths[index][0]))
def __len__(self):
return len(self.audiopaths)
class TextAudioCollate:
def __call__(self, batch):
batch = [b for b in batch if b is not None]
input_lengths, ids_sorted_decreasing = torch.sort(
torch.LongTensor([x[0].shape[1] for x in batch]),
dim=0, descending=True)
max_c_len = max([x[0].size(1) for x in batch])
max_wav_len = max([x[3].size(1) for x in batch])
lengths = torch.LongTensor(len(batch))
c_padded = torch.FloatTensor(len(batch), batch[0][0].shape[0], max_c_len)
f0_padded = torch.FloatTensor(len(batch), max_c_len)
spec_padded = torch.FloatTensor(len(batch), batch[0][2].shape[0], max_c_len)
wav_padded = torch.FloatTensor(len(batch), 1, max_wav_len)
spkids = torch.LongTensor(len(batch), 1)
uv_padded = torch.FloatTensor(len(batch), max_c_len)
volume_padded = torch.FloatTensor(len(batch), max_c_len)
c_padded.zero_()
spec_padded.zero_()
f0_padded.zero_()
wav_padded.zero_()
uv_padded.zero_()
volume_padded.zero_()
for i in range(len(ids_sorted_decreasing)):
row = batch[ids_sorted_decreasing[i]]
c = row[0]
c_padded[i, :, :c.size(1)] = c
lengths[i] = c.size(1)
f0 = row[1]
f0_padded[i, :f0.size(0)] = f0
spec = row[2]
spec_padded[i, :, :spec.size(1)] = spec
wav = row[3]
wav_padded[i, :, :wav.size(1)] = wav
spkids[i, 0] = row[4]
uv = row[5]
uv_padded[i, :uv.size(0)] = uv
volume = row[6]
if volume is not None:
volume_padded[i, :volume.size(0)] = volume
else :
volume_padded = None
return c_padded, f0_padded, spec_padded, wav_padded, spkids, lengths, uv_padded, volume_padded