ecapa_tdnn.py
6.7 KB
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
import torch
import torch.nn as nn
import torch.nn.functional as F
from typing import Optional, Tuple, List
class SEModule(nn.Module):
def __init__(self, channels, se_channels=128):
super().__init__()
self.se = nn.Sequential(
nn.Conv1d(channels, se_channels, kernel_size=1),
nn.ReLU(),
nn.BatchNorm1d(se_channels),
nn.Conv1d(se_channels, channels, kernel_size=1),
nn.Sigmoid(),
)
def forward(self, x):
return x * self.se(x)
class BandSplitBlock(nn.Module):
def __init__(self, n_mels: int, split_points: Optional[List[int]] = None, out_channels: int = 128):
super().__init__()
self.split_points = split_points or [16, 32, 64, 96, n_mels]
starts = [0] + self.split_points[:-1]
widths = [end - start for start, end in zip(starts, self.split_points)]
self.band_projs = nn.ModuleList(
[
nn.Sequential(
nn.Conv1d(width, out_channels, kernel_size=1),
nn.ReLU(),
nn.BatchNorm1d(out_channels),
)
for width in widths
]
)
self.fuse = nn.Sequential(
nn.Conv1d(out_channels * len(widths), out_channels * len(widths), kernel_size=1),
nn.ReLU(),
nn.BatchNorm1d(out_channels * len(widths)),
)
def forward(self, x):
starts = [0] + self.split_points[:-1]
bands = []
for proj, start, end in zip(self.band_projs, starts, self.split_points):
bands.append(proj(x[:, start:end, :]))
return self.fuse(torch.cat(bands, dim=1))
class Res2Block(nn.Module):
def __init__(self, channels, kernel_size=3, dilation=1, scale=8, se_channels=128):
super().__init__()
self.width = channels // scale
self.num_split = scale
self.convs = nn.ModuleList()
for _ in range(self.num_split):
self.convs.append(
nn.Sequential(
nn.Conv1d(
self.width,
self.width,
kernel_size,
padding=dilation * (kernel_size - 1) // 2,
dilation=dilation,
),
nn.ReLU(),
nn.BatchNorm1d(self.width),
)
)
self.conv1x1 = nn.Sequential(
nn.Conv1d(channels, channels, kernel_size=1),
nn.ReLU(),
nn.BatchNorm1d(channels),
)
self.se = SEModule(channels, se_channels)
def forward(self, x):
residual = x
split_x = torch.split(x, self.width, dim=1)
out = []
for i, (part, conv) in enumerate(zip(split_x, self.convs)):
if i == 0:
out.append(conv(part))
else:
out.append(conv(part + out[-1]))
x = torch.cat(out, dim=1)
x = self.conv1x1(x)
x = self.se(x)
return x + residual
class StatisticsPooling(nn.Module):
def forward(self, x):
mean = torch.mean(x, dim=2)
std = torch.sqrt(torch.var(x, dim=2, unbiased=False) + 1e-12)
return torch.cat([mean, std], dim=1)
class AAMSoftmax(nn.Module):
def __init__(self, in_features, out_features, m=0.3, s=30.0):
super().__init__()
self.weight = nn.Parameter(torch.FloatTensor(out_features, in_features))
nn.init.xavier_normal_(self.weight)
self.m = m
self.s = s
self.cos_m = torch.cos(torch.tensor(m))
self.sin_m = torch.sin(torch.tensor(m))
self.th = torch.cos(torch.tensor(torch.pi - m))
self.mm = torch.sin(torch.tensor(torch.pi - m)) * m
def forward(self, x, labels):
x = F.normalize(x, dim=1)
w = F.normalize(self.weight, dim=1)
cos_theta = F.linear(x, w)
sin_theta = torch.sqrt(1.0 - torch.clamp(cos_theta ** 2, 0, 1))
phi = cos_theta * self.cos_m - sin_theta * self.sin_m
phi = torch.where(cos_theta > self.th, phi, cos_theta - self.mm)
one_hot = F.one_hot(labels, num_classes=self.weight.size(0)).float()
output = (one_hot * phi) + ((1.0 - one_hot) * cos_theta)
output *= self.s
return output
class ECAPA_ACR(nn.Module):
def __init__(
self,
n_mels: int = 128,
embed_dim: int = 192,
channels: int = 512,
se_channels: int = 128,
res2net_scale: int = 8,
num_blocks: int = 3,
num_classes: Optional[int] = None,
aam_m: float = 0.3,
aam_s: float = 30.0,
use_band_split: bool = True,
band_split_channels: int = 128,
):
super().__init__()
self.embed_dim = embed_dim
front_channels = band_split_channels * 5 if use_band_split else channels
self.band_split = BandSplitBlock(n_mels=n_mels, out_channels=band_split_channels) if use_band_split else None
self.conv1 = nn.Sequential(
nn.Conv1d(front_channels, channels, kernel_size=5, stride=1, padding=2),
nn.ReLU(),
nn.BatchNorm1d(channels),
)
dilations = [1, 2, 3] if num_blocks == 3 else [d for d in range(1, num_blocks + 1)]
self.blocks = nn.ModuleList(
[
Res2Block(
channels=channels,
kernel_size=3,
dilation=d,
scale=res2net_scale,
se_channels=se_channels,
)
for d in dilations[:num_blocks]
]
)
in_channels = channels * num_blocks
self.mfa = nn.Sequential(
nn.Conv1d(in_channels, channels * 3, kernel_size=1),
nn.ReLU(),
nn.BatchNorm1d(channels * 3),
)
self.pooling = StatisticsPooling()
self.fc = nn.Linear(channels * 3 * 2, embed_dim)
self.bn = nn.BatchNorm1d(embed_dim, affine=False)
self.aam = AAMSoftmax(embed_dim, num_classes, m=aam_m, s=aam_s) if num_classes is not None else None
def forward(self, mel: torch.Tensor, labels: Optional[torch.Tensor] = None) -> Tuple[torch.Tensor, Optional[torch.Tensor]]:
x = self.band_split(mel) if self.band_split is not None else mel
x = self.conv1(x)
block_outputs = []
for block in self.blocks:
x = block(x)
block_outputs.append(x)
x = torch.cat(block_outputs, dim=1)
x = self.mfa(x)
x = self.pooling(x)
x = self.fc(x)
x = self.bn(x)
embedding = F.normalize(x, p=2, dim=1)
if labels is not None and self.aam is not None:
logits = self.aam(embedding, labels)
return embedding, logits
return embedding, None