File size: 6,910 Bytes
c3b58fa |
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 |
import math
from typing import Tuple, Optional
import torch
from torch import nn
from .attentions import MultiHeadAttention, FFN
from .norms import LayerNorm, WN
from .utils import sequence_mask
class Encoder(nn.Module):
def __init__(
self,
hidden_channels: int,
filter_channels: int,
n_heads: int,
n_layers: int,
kernel_size: int = 1,
p_dropout: float = 0.0,
window_size: int = 10,
):
super(Encoder, self).__init__()
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.window_size = window_size
self.drop = nn.Dropout(p_dropout)
self.attn_layers = nn.ModuleList()
self.norm_layers_1 = nn.ModuleList()
self.ffn_layers = nn.ModuleList()
self.norm_layers_2 = nn.ModuleList()
for _ in range(self.n_layers):
self.attn_layers.append(
MultiHeadAttention(
hidden_channels,
hidden_channels,
n_heads,
p_dropout=p_dropout,
window_size=window_size,
)
)
self.norm_layers_1.append(LayerNorm(hidden_channels))
self.ffn_layers.append(
FFN(
hidden_channels,
hidden_channels,
filter_channels,
kernel_size,
p_dropout=p_dropout,
)
)
self.norm_layers_2.append(LayerNorm(hidden_channels))
def __call__(self, x: torch.Tensor, x_mask: torch.Tensor) -> torch.Tensor:
return super().__call__(x, x_mask)
def forward(self, x: torch.Tensor, x_mask: torch.Tensor) -> torch.Tensor:
attn_mask = x_mask.unsqueeze(2) * x_mask.unsqueeze(-1)
x = x * x_mask
for attn, norm1, ffn, norm2 in zip(
self.attn_layers,
self.norm_layers_1,
self.ffn_layers,
self.norm_layers_2,
):
y = attn(x, x, attn_mask)
y = self.drop(y)
x = norm1(x + y)
y = ffn(x, x_mask)
y = self.drop(y)
x = norm2(x + y)
x = x * x_mask
return x
class TextEncoder(nn.Module):
def __init__(
self,
in_channels: int,
out_channels: int,
hidden_channels: int,
filter_channels: int,
n_heads: int,
n_layers: int,
kernel_size: int,
p_dropout: float,
f0: bool = True,
):
super(TextEncoder, self).__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 = float(p_dropout)
self.emb_phone = nn.Linear(in_channels, hidden_channels)
self.lrelu = nn.LeakyReLU(0.1, inplace=True)
if f0 == True:
self.emb_pitch = nn.Embedding(256, hidden_channels) # pitch 256
self.encoder = Encoder(
hidden_channels,
filter_channels,
n_heads,
n_layers,
kernel_size,
float(p_dropout),
)
self.proj = nn.Conv1d(hidden_channels, out_channels * 2, 1)
def __call__(
self,
phone: torch.Tensor,
pitch: torch.Tensor,
lengths: torch.Tensor,
skip_head: Optional[int] = None,
) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor]:
return super().__call__(
phone,
pitch,
lengths,
skip_head=skip_head,
)
def forward(
self,
phone: torch.Tensor,
pitch: torch.Tensor,
lengths: torch.Tensor,
skip_head: Optional[int] = None,
) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor]:
x = self.emb_phone(phone)
if pitch is not None:
x += self.emb_pitch(pitch)
x = x * math.sqrt(self.hidden_channels) # [b, t, h]
x = self.lrelu(x)
x = torch.transpose(x, 1, -1) # [b, h, t]
x_mask = torch.unsqueeze(
sequence_mask(lengths, x.size(2)),
1,
).to(x.dtype)
x = self.encoder(x * x_mask, x_mask)
if skip_head is not None:
head = int(skip_head)
x = x[:, :, head:]
x_mask = x_mask[:, :, head:]
stats: torch.Tensor = self.proj(x) * x_mask
m, logs = torch.split(stats, self.out_channels, dim=1)
return m, logs, x_mask
class PosteriorEncoder(nn.Module):
def __init__(
self,
in_channels: int,
out_channels: int,
hidden_channels: int,
kernel_size: int,
dilation_rate: int,
n_layers: int,
gin_channels=0,
):
super(PosteriorEncoder, self).__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 = WN(
hidden_channels,
kernel_size,
dilation_rate,
n_layers,
gin_channels=gin_channels,
)
self.proj = nn.Conv1d(hidden_channels, out_channels * 2, 1)
def __call__(
self, x: torch.Tensor, x_lengths: torch.Tensor, g: Optional[torch.Tensor] = None
) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor, torch.Tensor]:
return super().__call__(x, x_lengths, g=g)
def forward(
self, x: torch.Tensor, x_lengths: torch.Tensor, g: Optional[torch.Tensor] = None
) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor, torch.Tensor]:
x_mask = torch.unsqueeze(
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
def remove_weight_norm(self):
self.enc.remove_weight_norm()
def __prepare_scriptable__(self):
for hook in self.enc._forward_pre_hooks.values():
if (
hook.__module__ == "torch.nn.utils.weight_norm"
and hook.__class__.__name__ == "WeightNorm"
):
torch.nn.utils.remove_weight_norm(self.enc)
return self
|