-
Notifications
You must be signed in to change notification settings - Fork 12
/
Copy pathdataset.py
180 lines (144 loc) · 6.97 KB
/
dataset.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
from glob import glob
import os
import random
import numpy as np
import torch
import torch.utils.data
import torchaudio
import utils
import torchaudio.transforms as T
import random
"""Multi speaker version"""
class TestDataset(torch.utils.data.Dataset):
def __init__(self, audio_path, cfg, codec, all_in_mem: bool = False):
self.audiopaths = glob(os.path.join(audio_path, "**/*.wav"), recursive=True)
self.sampling_rate = cfg['data']['sampling_rate']
self.hop_length = cfg['data']['hop_length']
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):
audio, sampling_rate = torchaudio.load(filename)
audio = T.Resample(sampling_rate, self.sampling_rate)(audio)
spec = torch.load(filename.replace(".wav", ".spec.pt")).squeeze(0)
f0 = np.load(filename + ".f0.npy")
f0, uv = utils.interpolate_f0(f0)
f0 = torch.FloatTensor(f0)
uv = torch.FloatTensor(uv)
c = torch.load(filename+ ".soft.pt")
c = utils.repeat_expand_2d(c.squeeze(0), f0.shape[0])
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.shape[1]-lmin * self.hop_length) < 3 * self.hop_length
spec, c, f0, uv = spec[:, :lmin], c[:, :lmin], f0[:lmin], uv[:lmin]
audio = audio[:, :lmin * self.hop_length]
return c.detach(), f0.detach(), spec.detach(), audio.detach(), uv.detach()
def __getitem__(self, index):
return *self.get_audio(self.audiopaths[index]), *self.get_audio(self.audiopaths[(index+4)%self.__len__()])
def __len__(self):
return len(self.audiopaths)
class NS2VCDataset(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, audio_path, cfg, codec, all_in_mem: bool = False):
self.audiopaths = glob(os.path.join(audio_path, "**/*.wav"), recursive=True)
self.sampling_rate = cfg['data']['sampling_rate']
self.hop_length = cfg['data']['hop_length']
# self.codec = codec
# 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):
audio, sampling_rate = torchaudio.load(filename)
audio = T.Resample(sampling_rate, self.sampling_rate)(audio)
spec = torch.load(filename.replace(".wav", ".spec.pt")).squeeze(0)
f0 = np.load(filename + ".f0.npy")
f0, uv = utils.interpolate_f0(f0)
f0 = torch.FloatTensor(f0)
uv = torch.FloatTensor(uv)
c = torch.load(filename+ ".soft.pt")
c = utils.repeat_expand_2d(c.squeeze(0), f0.shape[0])
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.shape[1]-lmin * self.hop_length) < 3 * self.hop_length
spec, c, f0, uv = spec[:, :lmin], c[:, :lmin], f0[:lmin], uv[:lmin]
audio = audio[:, :lmin * self.hop_length]
return c.detach(), f0.detach(), spec.detach(), audio.detach(), uv.detach()
def random_slice(self, c, f0, spec, audio, uv):
if spec.shape[1] < 30:
print("skip too short audio")
return None
if spec.shape[1] > 400:
start = random.randint(0, spec.shape[1]-400)
end = start + 400
spec, c, f0, uv = spec[:, start:end], c[:, start:end], f0[start:end], uv[start:end]
audio = audio[:, start * self.hop_length : end * self.hop_length]
len_spec = spec.shape[1]
l = random.randint(int(len_spec//3), int(len_spec//3*2))
u = random.randint(0, len_spec-l)
v = u + l
refer = spec[:, u:v]
c = torch.cat([c[:, :u], c[:, v:]], dim=-1)
f0 = torch.cat([f0[:u], f0[v:]], dim=-1)
spec = torch.cat([spec[:, :u], spec[:, v:]], dim=-1)
uv = torch.cat([uv[:u], uv[v:]], dim=-1)
audio = torch.cat([audio[:, :u * self.hop_length], audio[:, v * self.hop_length:]], dim=-1)
assert c.shape[1] != 0
assert refer.shape[1] != 0
return refer, c, f0, spec, audio, uv
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]))
# print(1)
def __len__(self):
return len(self.audiopaths)
class TextAudioCollate:
def __call__(self, batch):
hop_length = 320
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)
# refer, c, f0, spec, audio, uv
max_refer_len = max([x[0].size(1) for x in batch])
max_c_len = max([x[1].size(1) for x in batch])
max_wav_len = max([x[4].size(1) for x in batch])
lengths = torch.LongTensor(len(batch))
refer_lengths = torch.LongTensor(len(batch))
contentvec_dim = batch[0][1].shape[0]
spec_dim = batch[0][3].shape[0]
c_padded = torch.FloatTensor(len(batch), contentvec_dim, max_c_len+1)
f0_padded = torch.FloatTensor(len(batch), max_c_len+1)
spec_padded = torch.FloatTensor(len(batch), spec_dim, max_c_len+1)
refer_padded = torch.FloatTensor(len(batch), spec_dim, max_refer_len+1)
wav_padded = torch.FloatTensor(len(batch), 1, max_wav_len+1)
uv_padded = torch.FloatTensor(len(batch), max_c_len+1)
c_padded.zero_()
spec_padded.zero_()
refer_padded.zero_()
f0_padded.zero_()
wav_padded.zero_()
uv_padded.zero_()
for i in range(len(ids_sorted_decreasing)):
row = batch[ids_sorted_decreasing[i]]
# refer, c, f0, spec, audio, uv
len_refer = row[0].size(1)
len_contentvec = row[1].size(1)
len_wav = row[4].size(1)
lengths[i] = len_contentvec
refer_lengths[i] = len_refer
refer_padded[i, :, :len_refer] = row[0][:]
c_padded[i, :, :len_contentvec] = row[1][:]
f0_padded[i, :len_contentvec] = row[2][:]
spec_padded[i, :, :len_contentvec] = row[3][:]
wav_padded[i, :, :len_wav] = row[4][:]
uv_padded[i, :len_contentvec] = row[5][:]
return c_padded, refer_padded, f0_padded, spec_padded, wav_padded, lengths, refer_lengths, uv_padded