File size: 11,920 Bytes
a726cc5
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.nn.utils import weight_norm, remove_weight_norm
from torch.nn import Conv1d, ConvTranspose1d

LRELU_SLOPE = 0.1
alpha = 1.0

def get_padding(kernel_size, dilation=1):
    return int((kernel_size*dilation - dilation)/2)

def init_weights(m, mean=0.0, std=0.01):
    classname = m.__class__.__name__
    if classname.find("Conv") != -1:
        m.weight.data.normal_(mean, std)

class ResBlock1(torch.nn.Module):
    def __init__(self, h, channels, kernel_size=3, dilation=(1, 3, 5)):
        super(ResBlock1, self).__init__()
        self.h = h
        self.convs1 = nn.ModuleList([
            weight_norm(Conv1d(channels, channels, kernel_size, 1, dilation=dilation[0],
                               padding=get_padding(kernel_size, dilation[0]))),
            weight_norm(Conv1d(channels, channels, kernel_size, 1, dilation=dilation[1],
                               padding=get_padding(kernel_size, dilation[1]))),
            weight_norm(Conv1d(channels, channels, kernel_size, 1, dilation=dilation[2],
                               padding=get_padding(kernel_size, dilation[2])))
        ])
        self.convs1.apply(init_weights)

        self.convs2 = nn.ModuleList([
            weight_norm(Conv1d(channels, channels, kernel_size, 1, dilation=1,
                               padding=get_padding(kernel_size, 1))),
            weight_norm(Conv1d(channels, channels, kernel_size, 1, dilation=1,
                               padding=get_padding(kernel_size, 1))),
            weight_norm(Conv1d(channels, channels, kernel_size, 1, dilation=1,
                               padding=get_padding(kernel_size, 1)))
        ])
        self.convs2.apply(init_weights)
        self.num_layers = len(self.convs1) + len(self.convs2) # total number of conv layers
        self.activations = nn.ModuleList([nn.LeakyReLU(LRELU_SLOPE) for _ in range(self.num_layers)])


    def forward(self, x):
        acts1, acts2 = self.activations[::2], self.activations[1::2]
        for c1, c2,a1,a2 in zip(self.convs1, self.convs2,acts1,acts2):
            xt = a1(x)
            xt = c1(xt)
            xt = a2(xt)
            xt = c2(xt)
            x = xt + x
        return x

    def remove_weight_norm(self):
        for l in self.convs1:
            remove_weight_norm(l)
        for l in self.convs2:
            remove_weight_norm(l)


class Encoder(torch.nn.Module):
    def __init__(self, h):
        super(Encoder, self).__init__()
        self.n_filters = h.en_filters
        self.vq_dim = h.vq_dim
        self.num_kernels = len(h.resblock_kernel_sizes)
        self.num_upsamples = len(h.upsample_rates)
        self.upsample_initial_channel = self.n_filters * ( 2**self.num_upsamples )
        self.conv_pre = weight_norm(Conv1d(h.channel, self.n_filters, 7, 1, padding=3))
        self.normalize = nn.ModuleList()
        resblock = ResBlock1 

        self.ups = nn.ModuleList()
        for i, (u, k) in enumerate(list(reversed(list(zip(h.upsample_rates, h.upsample_kernel_sizes))))):
            self.ups.append(weight_norm(
                Conv1d(self.n_filters*(2**i), self.n_filters*(2**(i+1)),
                       k, u,
                       padding=((k-u)//2)
                )))
        self.resblocks = nn.ModuleList()
        ch = 1
        for i in range(len(self.ups)):
            ch = self.n_filters*(2**(i+1))
            for j, (k, d) in enumerate(
                    zip(
                        list(reversed(h.resblock_kernel_sizes)),
                        list(reversed(h.resblock_dilation_sizes))
                    )
            ):
                self.resblocks.append(resblock(h, ch, k, d))
                self.normalize.append(torch.nn.LayerNorm([ch],eps=1e-6,elementwise_affine=True))
        
        self.activation_post = nn.LeakyReLU(LRELU_SLOPE)
        self.conv_post = Conv1d(ch, self.vq_dim, 3, 1, padding=1)
        self.ups.apply(init_weights)
        self.conv_post.apply(init_weights)

    def forward(self, x):
        x = self.conv_pre(x)
        for i in range(self.num_upsamples):
            x = self.ups[i](x)
            xs = None
            for j in range(self.num_kernels):
                if xs is None:
                    xs = self.resblocks[i*self.num_kernels+j](x)
                    xs = self.normalize[i*self.num_kernels+j](xs.transpose(1,2)).transpose(1,2)
                else:
                    xs += self.resblocks[i*self.num_kernels+j](x)
                    xs = self.normalize[i*self.num_kernels+j](xs.transpose(1,2)).transpose(1,2)
            x = xs / self.num_kernels
        x = self.activation_post(x)
        x = self.conv_post(x)
        return x

    def remove_weight_norm(self):
        print('Removing weight norm...')
        for l in self.ups:
            remove_weight_norm(l)
        for l in self.resblocks:
            l.remove_weight_norm()
        remove_weight_norm(self.conv_pre)

class Quantizer_module(torch.nn.Module):
    def __init__(self, n_e, e_dim):
        super(Quantizer_module, self).__init__()
        self.embedding = nn.Embedding(n_e, e_dim)
        self.embedding.weight.data.uniform_(-1.0 / n_e, 1.0 / n_e)
        self.target = torch.arange(0,n_e)

    def forward(self, x, idx=0):
        loss=torch.Tensor([0.0])
        d = torch.sum(x ** 2, 1, keepdim=True) + torch.sum(self.embedding.weight ** 2, 1) \
            - 2 * torch.matmul(x, self.embedding.weight.T)
        min_indicies = torch.argmin(d, 1)
        z_q = self.embedding(min_indicies)
        embed_vec = self.embedding.weight
        embed_dis = torch.mm(embed_vec , embed_vec.T)*3
        self.target = torch.arange(0,embed_vec.shape[0]).to(x.device)
        loss = F.cross_entropy(embed_dis,self.target)*(idx==0)
        return z_q, min_indicies,loss

class Quantizer(torch.nn.Module):
    def __init__(self, h):
        super(Quantizer, self).__init__()
        assert h.vq_dim % h.n_code_groups == 0
        self.lm_offset = 0
        self.lm_states = None
        self.vq_dim = h.vq_dim
        self.residul_layer = h.n_q
        self.n_code_groups = h.n_code_groups
        self.quantizer_modules = nn.ModuleList()
        for i in range(self.residul_layer):
            self.quantizer_modules.append(nn.ModuleList([
                Quantizer_module(h.n_codes, self.vq_dim // h.n_code_groups) for _ in range(h.n_code_groups)
            ]))
        self.h = h
        self.codebook_loss_lambda = self.h.codebook_loss_lambda     # e.g., 1
        self.commitment_loss_lambda = self.h.commitment_loss_lambda # e.g., 0.25
        

    def for_one_step(self, xin, idx):
        xin = xin.transpose(1, 2)
        x = xin.reshape(-1, self.vq_dim)
        x = torch.split(x, self.vq_dim // self.h.n_code_groups, dim=-1)
        min_indicies = []
        z_q = []
        all_losses = []
        for _x, m in zip(x, self.quantizer_modules[idx]):
            _z_q, _min_indicies,_loss = m(_x,idx)
            all_losses.append(_loss)
            z_q.append(_z_q)
            min_indicies.append(_min_indicies) 
        z_q = torch.cat(z_q, -1).reshape(xin.shape)
        z_q = z_q.transpose(1, 2)
        all_losses = torch.stack(all_losses)
        loss = torch.mean(all_losses)
        return z_q, min_indicies, loss
        
    
    def forward(self, xin,bw=-1,mask_id=None):
        quantized_out = 0.0
        residual = xin
        all_losses = []
        all_indices = []
        if bw<=0:
            bw = self.residul_layer
        for i in range(bw):
            quantized,  indices, e_loss = self.for_one_step(residual, i) # 
            if mask_id is not None:
                mask = (
                    torch.full([xin.shape[0],xin.shape[2],1], fill_value=i, device=xin.device) < mask_id.unsqueeze(2) + 1
                )
                mask = mask.repeat(1,1,xin.shape[1]).transpose(1,2)
            if mask_id is not None:
                loss = 0.1 * e_loss + self.codebook_loss_lambda * torch.mean((quantized - residual.detach()) ** 2 * mask) \
                    + self.commitment_loss_lambda * torch.mean((quantized.detach() - residual) ** 2 * mask ) 
            else:
                loss = 0.1 * e_loss \
                    +  self.codebook_loss_lambda * torch.mean((quantized - residual.detach()) ** 2 ) \
                    + self.commitment_loss_lambda * torch.mean((quantized.detach() - residual) ** 2  ) 
                
            quantized = residual + (quantized - residual).detach()
            residual = residual - quantized
            if mask_id is not None:
                quantized_out = quantized_out + quantized * mask
            else:
                quantized_out = quantized_out + quantized
            all_indices.extend(indices) # 
            all_losses.append(loss)
        all_losses = torch.stack(all_losses)
        loss = torch.mean(all_losses)
        return quantized_out, loss, all_indices
    
    def embed(self, x , bw=-1):
        quantized_out = torch.tensor(0.0, device=x.device)
        x = torch.split(x, 1, 2) 
        if bw <= 0 or bw > self.residul_layer:
            bw = self.residul_layer
        for i in range(bw):
            ret = []
            for j in range(self.n_code_groups):
                q = x[j+self.n_code_groups*i]
                embed = self.quantizer_modules[i][j]
                q = embed.embedding(q.squeeze(-1))
                ret.append(q)
            ret = torch.cat(ret, -1)
            quantized_out = quantized_out + ret
        return quantized_out.transpose(1, 2)


class Generator(torch.nn.Module):
    def __init__(self, h):
        super(Generator, self).__init__()
        self.h = h
        self.n_filters = h.de_filters
        self.vq_dim = h.vq_dim
        self.num_kernels = len(h.resblock_kernel_sizes)
        self.num_upsamples = len(h.upsample_rates)
        self.upsample_initial_channel = self.n_filters * ( 2**self.num_upsamples )
        self.conv_pre = weight_norm(Conv1d(self.vq_dim, self.upsample_initial_channel, 7, 1, padding=3))
        resblock = ResBlock1
        

        self.norm = nn.Identity()

        self.ups = nn.ModuleList()
        for i, (u, k) in enumerate(zip(h.upsample_rates, h.upsample_kernel_sizes)):
            self.ups.append(weight_norm(
                ConvTranspose1d(
                    self.upsample_initial_channel//(2**i), self.upsample_initial_channel//(2**(i+1)),
                                k, u,
                                padding=(k - u )//2,
                )
            ))
        ch = 1
        self.resblocks = nn.ModuleList()
        for i in range(len(self.ups)):
            ch = self.upsample_initial_channel//(2**(i+1))
            for j, (k, d) in enumerate(zip(h.resblock_kernel_sizes, h.resblock_dilation_sizes)):
                self.resblocks.append(resblock(h, ch, k, d))
        
        
        self.activation_post = nn.LeakyReLU(LRELU_SLOPE)
        self.conv_post = weight_norm(Conv1d(ch, h.channel, 7, 1, padding=3))
        self.ups.apply(init_weights)
        self.conv_post.apply(init_weights)

    def forward(self, x):
        x = self.norm(x)
        x = self.conv_pre(x)
        
        for i in range(self.num_upsamples):
            x = self.ups[i](x)
            xs = None
            for j in range(self.num_kernels):
                if xs is None:
                    xs = self.resblocks[i*self.num_kernels+j](x)
                else:
                    xs += self.resblocks[i*self.num_kernels+j](x)
            x = xs / self.num_kernels
        x = self.activation_post(x)
        x = self.conv_post(x)
        x = torch.tanh(x)

        return x

    def remove_weight_norm(self):
        print('Removing weight norm...')
        for l in self.ups:
            remove_weight_norm(l)
        for l in self.resblocks:
            l.remove_weight_norm()
        remove_weight_norm(self.conv_pre)
        remove_weight_norm(self.conv_post)