File size: 14,495 Bytes
5a030e1
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
import copy
import math
import torch
from torch import nn
from torch.nn import functional as F

import modules.attentions as attentions
import modules.commons as commons
import modules.modules as modules

from torch.nn import Conv1d, ConvTranspose1d, AvgPool1d, Conv2d
from torch.nn.utils import weight_norm, remove_weight_norm, spectral_norm

import utils
from modules.commons import init_weights, get_padding
from vdecoder.hifigan.models import Generator
from utils import f0_to_coarse

class ResidualCouplingBlock(nn.Module):
  def __init__(self,
      channels,
      hidden_channels,
      kernel_size,
      dilation_rate,
      n_layers,
      n_flows=4,
      gin_channels=0):
    super().__init__()
    self.channels = channels
    self.hidden_channels = hidden_channels
    self.kernel_size = kernel_size
    self.dilation_rate = dilation_rate
    self.n_layers = n_layers
    self.n_flows = n_flows
    self.gin_channels = gin_channels

    self.flows = nn.ModuleList()
    for i in range(n_flows):
      self.flows.append(modules.ResidualCouplingLayer(channels, hidden_channels, kernel_size, dilation_rate, n_layers, gin_channels=gin_channels, mean_only=True))
      self.flows.append(modules.Flip())

  def forward(self, x, x_mask, g=None, reverse=False):
    if not reverse:
      for flow in self.flows:
        x, _ = flow(x, x_mask, g=g, reverse=reverse)
    else:
      for flow in reversed(self.flows):
        x = flow(x, x_mask, g=g, reverse=reverse)
    return x


class Encoder(nn.Module):
  def __init__(self,
      in_channels,
      out_channels,
      hidden_channels,
      kernel_size,
      dilation_rate,
      n_layers,
      gin_channels=0):
    super().__init__()
    self.in_channels = in_channels
    self.out_channels = out_channels
    self.hidden_channels = hidden_channels
    self.kernel_size = kernel_size
    self.dilation_rate = dilation_rate
    self.n_layers = n_layers
    self.gin_channels = gin_channels

    self.pre = nn.Conv1d(in_channels, hidden_channels, 1)
    self.enc = modules.WN(hidden_channels, kernel_size, dilation_rate, n_layers, gin_channels=gin_channels)
    self.proj = nn.Conv1d(hidden_channels, out_channels * 2, 1)

  def forward(self, x, x_lengths, g=None):
    # print(x.shape,x_lengths.shape)
    x_mask = torch.unsqueeze(commons.sequence_mask(x_lengths, x.size(2)), 1).to(x.dtype)
    x = self.pre(x) * x_mask
    x = self.enc(x, x_mask, g=g)
    stats = self.proj(x) * x_mask
    m, logs = torch.split(stats, self.out_channels, dim=1)
    z = (m + torch.randn_like(m) * torch.exp(logs)) * x_mask
    return z, m, logs, x_mask


class TextEncoder(nn.Module):
  def __init__(self,
      out_channels,
      hidden_channels,
      kernel_size,
      n_layers,
      gin_channels=0,
      filter_channels=None,
      n_heads=None,
      p_dropout=None):
    super().__init__()
    self.out_channels = out_channels
    self.hidden_channels = hidden_channels
    self.kernel_size = kernel_size
    self.n_layers = n_layers
    self.gin_channels = gin_channels
    self.proj = nn.Conv1d(hidden_channels, out_channels * 2, 1)
    self.f0_emb = nn.Embedding(256, hidden_channels)

    self.enc_ =  attentions.Encoder(
        hidden_channels,
        filter_channels,
        n_heads,
        n_layers,
        kernel_size,
        p_dropout)

  def forward(self, x, x_mask, f0=None, noice_scale=1):
    x = x + self.f0_emb(f0).transpose(1,2)
    x = self.enc_(x * x_mask, x_mask)
    stats = self.proj(x) * x_mask
    m, logs = torch.split(stats, self.out_channels, dim=1)
    z = (m + torch.randn_like(m) * torch.exp(logs) * noice_scale) * x_mask

    return z, m, logs, x_mask



class DiscriminatorP(torch.nn.Module):
    def __init__(self, period, kernel_size=5, stride=3, use_spectral_norm=False):
        super(DiscriminatorP, self).__init__()
        self.period = period
        self.use_spectral_norm = use_spectral_norm
        norm_f = weight_norm if use_spectral_norm == False else spectral_norm
        self.convs = nn.ModuleList([
            norm_f(Conv2d(1, 32, (kernel_size, 1), (stride, 1), padding=(get_padding(kernel_size, 1), 0))),
            norm_f(Conv2d(32, 128, (kernel_size, 1), (stride, 1), padding=(get_padding(kernel_size, 1), 0))),
            norm_f(Conv2d(128, 512, (kernel_size, 1), (stride, 1), padding=(get_padding(kernel_size, 1), 0))),
            norm_f(Conv2d(512, 1024, (kernel_size, 1), (stride, 1), padding=(get_padding(kernel_size, 1), 0))),
            norm_f(Conv2d(1024, 1024, (kernel_size, 1), 1, padding=(get_padding(kernel_size, 1), 0))),
        ])
        self.conv_post = norm_f(Conv2d(1024, 1, (3, 1), 1, padding=(1, 0)))

    def forward(self, x):
        fmap = []

        # 1d to 2d
        b, c, t = x.shape
        if t % self.period != 0: # pad first
            n_pad = self.period - (t % self.period)
            x = F.pad(x, (0, n_pad), "reflect")
            t = t + n_pad
        x = x.view(b, c, t // self.period, self.period)

        for l in self.convs:
            x = l(x)
            x = F.leaky_relu(x, modules.LRELU_SLOPE)
            fmap.append(x)
        x = self.conv_post(x)
        fmap.append(x)
        x = torch.flatten(x, 1, -1)

        return x, fmap


class DiscriminatorS(torch.nn.Module):
    def __init__(self, use_spectral_norm=False):
        super(DiscriminatorS, self).__init__()
        norm_f = weight_norm if use_spectral_norm == False else spectral_norm
        self.convs = nn.ModuleList([
            norm_f(Conv1d(1, 16, 15, 1, padding=7)),
            norm_f(Conv1d(16, 64, 41, 4, groups=4, padding=20)),
            norm_f(Conv1d(64, 256, 41, 4, groups=16, padding=20)),
            norm_f(Conv1d(256, 1024, 41, 4, groups=64, padding=20)),
            norm_f(Conv1d(1024, 1024, 41, 4, groups=256, padding=20)),
            norm_f(Conv1d(1024, 1024, 5, 1, padding=2)),
        ])
        self.conv_post = norm_f(Conv1d(1024, 1, 3, 1, padding=1))

    def forward(self, x):
        fmap = []

        for l in self.convs:
            x = l(x)
            x = F.leaky_relu(x, modules.LRELU_SLOPE)
            fmap.append(x)
        x = self.conv_post(x)
        fmap.append(x)
        x = torch.flatten(x, 1, -1)

        return x, fmap


class MultiPeriodDiscriminator(torch.nn.Module):
    def __init__(self, use_spectral_norm=False):
        super(MultiPeriodDiscriminator, self).__init__()
        periods = [2,3,5,7,11]

        discs = [DiscriminatorS(use_spectral_norm=use_spectral_norm)]
        discs = discs + [DiscriminatorP(i, use_spectral_norm=use_spectral_norm) for i in periods]
        self.discriminators = nn.ModuleList(discs)

    def forward(self, y, y_hat):
        y_d_rs = []
        y_d_gs = []
        fmap_rs = []
        fmap_gs = []
        for i, d in enumerate(self.discriminators):
            y_d_r, fmap_r = d(y)
            y_d_g, fmap_g = d(y_hat)
            y_d_rs.append(y_d_r)
            y_d_gs.append(y_d_g)
            fmap_rs.append(fmap_r)
            fmap_gs.append(fmap_g)

        return y_d_rs, y_d_gs, fmap_rs, fmap_gs


class SpeakerEncoder(torch.nn.Module):
    def __init__(self, mel_n_channels=80, model_num_layers=3, model_hidden_size=256, model_embedding_size=256):
        super(SpeakerEncoder, self).__init__()
        self.lstm = nn.LSTM(mel_n_channels, model_hidden_size, model_num_layers, batch_first=True)
        self.linear = nn.Linear(model_hidden_size, model_embedding_size)
        self.relu = nn.ReLU()

    def forward(self, mels):
        self.lstm.flatten_parameters()
        _, (hidden, _) = self.lstm(mels)
        embeds_raw = self.relu(self.linear(hidden[-1]))
        return embeds_raw / torch.norm(embeds_raw, dim=1, keepdim=True)

    def compute_partial_slices(self, total_frames, partial_frames, partial_hop):
        mel_slices = []
        for i in range(0, total_frames-partial_frames, partial_hop):
            mel_range = torch.arange(i, i+partial_frames)
            mel_slices.append(mel_range)

        return mel_slices

    def embed_utterance(self, mel, partial_frames=128, partial_hop=64):
        mel_len = mel.size(1)
        last_mel = mel[:,-partial_frames:]

        if mel_len > partial_frames:
            mel_slices = self.compute_partial_slices(mel_len, partial_frames, partial_hop)
            mels = list(mel[:,s] for s in mel_slices)
            mels.append(last_mel)
            mels = torch.stack(tuple(mels), 0).squeeze(1)

            with torch.no_grad():
                partial_embeds = self(mels)
            embed = torch.mean(partial_embeds, axis=0).unsqueeze(0)
            #embed = embed / torch.linalg.norm(embed, 2)
        else:
            with torch.no_grad():
                embed = self(last_mel)

        return embed

class F0Decoder(nn.Module):
    def __init__(self,
                 out_channels,
                 hidden_channels,
                 filter_channels,
                 n_heads,
                 n_layers,
                 kernel_size,
                 p_dropout,
                 spk_channels=0):
        super().__init__()
        self.out_channels = out_channels
        self.hidden_channels = hidden_channels
        self.filter_channels = filter_channels
        self.n_heads = n_heads
        self.n_layers = n_layers
        self.kernel_size = kernel_size
        self.p_dropout = p_dropout
        self.spk_channels = spk_channels

        self.prenet = nn.Conv1d(hidden_channels, hidden_channels, 3, padding=1)
        self.decoder = attentions.FFT(
            hidden_channels,
            filter_channels,
            n_heads,
            n_layers,
            kernel_size,
            p_dropout)
        self.proj = nn.Conv1d(hidden_channels, out_channels, 1)
        self.f0_prenet = nn.Conv1d(1, hidden_channels , 3, padding=1)
        self.cond = nn.Conv1d(spk_channels, hidden_channels, 1)

    def forward(self, x, norm_f0, x_mask, spk_emb=None):
        x = torch.detach(x)
        if (spk_emb is not None):
            x = x + self.cond(spk_emb)
        x += self.f0_prenet(norm_f0)
        x = self.prenet(x) * x_mask
        x = self.decoder(x * x_mask, x_mask)
        x = self.proj(x) * x_mask
        return x


class SynthesizerTrn(nn.Module):
  """
  Synthesizer for Training
  """

  def __init__(self,
    spec_channels,
    segment_size,
    inter_channels,
    hidden_channels,
    filter_channels,
    n_heads,
    n_layers,
    kernel_size,
    p_dropout,
    resblock,
    resblock_kernel_sizes,
    resblock_dilation_sizes,
    upsample_rates,
    upsample_initial_channel,
    upsample_kernel_sizes,
    gin_channels,
    ssl_dim,
    n_speakers,
    sampling_rate=44100,
    **kwargs):

    super().__init__()
    self.spec_channels = spec_channels
    self.inter_channels = inter_channels
    self.hidden_channels = hidden_channels
    self.filter_channels = filter_channels
    self.n_heads = n_heads
    self.n_layers = n_layers
    self.kernel_size = kernel_size
    self.p_dropout = p_dropout
    self.resblock = resblock
    self.resblock_kernel_sizes = resblock_kernel_sizes
    self.resblock_dilation_sizes = resblock_dilation_sizes
    self.upsample_rates = upsample_rates
    self.upsample_initial_channel = upsample_initial_channel
    self.upsample_kernel_sizes = upsample_kernel_sizes
    self.segment_size = segment_size
    self.gin_channels = gin_channels
    self.ssl_dim = ssl_dim
    self.emb_g = nn.Embedding(n_speakers, gin_channels)

    self.pre = nn.Conv1d(ssl_dim, hidden_channels, kernel_size=5, padding=2)

    self.enc_p = TextEncoder(
        inter_channels,
        hidden_channels,
        filter_channels=filter_channels,
        n_heads=n_heads,
        n_layers=n_layers,
        kernel_size=kernel_size,
        p_dropout=p_dropout
    )
    hps = {
        "sampling_rate": sampling_rate,
        "inter_channels": inter_channels,
        "resblock": resblock,
        "resblock_kernel_sizes": resblock_kernel_sizes,
        "resblock_dilation_sizes": resblock_dilation_sizes,
        "upsample_rates": upsample_rates,
        "upsample_initial_channel": upsample_initial_channel,
        "upsample_kernel_sizes": upsample_kernel_sizes,
        "gin_channels": gin_channels,
    }
    self.dec = Generator(h=hps)
    self.enc_q = Encoder(spec_channels, inter_channels, hidden_channels, 5, 1, 16, gin_channels=gin_channels)
    self.flow = ResidualCouplingBlock(inter_channels, hidden_channels, 5, 1, 4, gin_channels=gin_channels)
    self.f0_decoder = F0Decoder(
        1,
        hidden_channels,
        filter_channels,
        n_heads,
        n_layers,
        kernel_size,
        p_dropout,
        spk_channels=gin_channels
    )
    self.emb_uv = nn.Embedding(2, hidden_channels)

  def forward(self, c, f0, uv, spec, g=None, c_lengths=None, spec_lengths=None):
    g = self.emb_g(g).transpose(1,2)
    # ssl prenet
    x_mask = torch.unsqueeze(commons.sequence_mask(c_lengths, c.size(2)), 1).to(c.dtype)
    x = self.pre(c) * x_mask + self.emb_uv(uv.long()).transpose(1,2)

    # f0 predict
    lf0 = 2595. * torch.log10(1. + f0.unsqueeze(1) / 700.) / 500
    norm_lf0 = utils.normalize_f0(lf0, x_mask, uv)
    pred_lf0 = self.f0_decoder(x, norm_lf0, x_mask, spk_emb=g)

    # encoder
    z_ptemp, m_p, logs_p, _ = self.enc_p(x, x_mask, f0=f0_to_coarse(f0))
    z, m_q, logs_q, spec_mask = self.enc_q(spec, spec_lengths, g=g) 

    # flow
    z_p = self.flow(z, spec_mask, g=g)
    z_slice, pitch_slice, ids_slice = commons.rand_slice_segments_with_pitch(z, f0, spec_lengths, self.segment_size)

    # nsf decoder
    o = self.dec(z_slice, g=g, f0=pitch_slice)

    return o, ids_slice, spec_mask, (z, z_p, m_p, logs_p, m_q, logs_q), pred_lf0, norm_lf0, lf0

  def infer(self, c, f0, uv, g=None, noice_scale=0.35, predict_f0=False):
    c_lengths = (torch.ones(c.size(0)) * c.size(-1)).to(c.device)
    g = self.emb_g(g).transpose(1,2)
    x_mask = torch.unsqueeze(commons.sequence_mask(c_lengths, c.size(2)), 1).to(c.dtype)
    x = self.pre(c) * x_mask + self.emb_uv(uv.long()).transpose(1,2)

    if predict_f0:
        lf0 = 2595. * torch.log10(1. + f0.unsqueeze(1) / 700.) / 500
        norm_lf0 = utils.normalize_f0(lf0, x_mask, uv, random_scale=False)
        pred_lf0 = self.f0_decoder(x, norm_lf0, x_mask, spk_emb=g)
        f0 = (700 * (torch.pow(10, pred_lf0 * 500 / 2595) - 1)).squeeze(1)

    z_p, m_p, logs_p, c_mask = self.enc_p(x, x_mask, f0=f0_to_coarse(f0), noice_scale=noice_scale)
    z = self.flow(z_p, c_mask, g=g, reverse=True)
    o = self.dec(z * c_mask, g=g, f0=f0)
    return o