File size: 8,485 Bytes
9d61c9b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
from typing import Optional, Tuple

import numpy as np
import torch
from torch import Tensor, nn


def maximum_path(
    value: Tensor,
    mask: Tensor,
    max_neg_val: Optional[float] = None,
):
    """Monotonic alignment search algorithm
    Numpy-friendly version. It's about 4 times faster than torch version.
    value: [b, t_x, t_y]
    mask: [b, t_x, t_y]
    """
    if max_neg_val is None:
        max_neg_val = -np.inf  # Patch for Sphinx complaint
    value = value * mask

    device = value.device
    dtype = value.dtype
    value = value.cpu().detach().numpy()
    mask = mask.cpu().detach().numpy().astype(bool)

    b, t_x, t_y = value.shape
    direction = np.zeros(value.shape, dtype=np.int64)
    v = np.zeros((b, t_x), dtype=np.float32)
    x_range = np.arange(t_x, dtype=np.float32).reshape(1, -1)
    for j in range(t_y):
        v0 = np.pad(v, [[0, 0], [1, 0]], mode="constant", constant_values=max_neg_val)[
            :,
            :-1,
        ]
        v1 = v
        max_mask = v1 >= v0
        v_max = np.where(max_mask, v1, v0)
        direction[:, :, j] = max_mask

        index_mask = x_range <= j
        v = np.where(index_mask, v_max + value[:, :, j], max_neg_val)
    direction = np.where(mask, direction, 1)

    path = np.zeros(value.shape, dtype=np.float32)
    index = mask[:, :, 0].sum(1).astype(np.int64) - 1  # type: ignore
    index_range = np.arange(b)
    for j in reversed(range(t_y)):
        path[index_range, index, j] = 1
        index = index + direction[index_range, index, j] - 1
    path = path * mask.astype(np.float32)  # type: ignore
    path = torch.from_numpy(path).to(device=device, dtype=dtype)
    return path


class AlignmentNetwork(torch.nn.Module):
    r"""Aligner Network for learning alignment between the input text and the model output with Gaussian Attention.

    The architecture is as follows:
    query -> conv1d -> relu -> conv1d -> relu -> conv1d -> L2_dist -> softmax -> alignment
    key   -> conv1d -> relu -> conv1d -----------------------^

    Args:
        in_query_channels (int): Number of channels in the query network.
        in_key_channels (int): Number of channels in the key network.
        attn_channels (int): Number of inner channels in the attention layers.
        temperature (float, optional): Temperature for the softmax. Defaults to 0.0005.
    """

    def __init__(
        self,
        in_query_channels: int,
        in_key_channels: int,
        attn_channels: int,
        temperature: float = 0.0005,
    ):
        super().__init__()
        self.temperature = temperature
        self.softmax = torch.nn.Softmax(dim=3)
        self.log_softmax = torch.nn.LogSoftmax(dim=3)

        self.key_layer = nn.Sequential(
            nn.Conv1d(
                in_key_channels,
                in_key_channels * 2,
                kernel_size=3,
                padding=1,
                bias=True,
            ),
            torch.nn.ReLU(),
            nn.Conv1d(
                in_key_channels * 2,
                attn_channels,
                kernel_size=1,
                padding=0,
                bias=True,
            ),
        )

        self.query_layer = nn.Sequential(
            nn.Conv1d(
                in_query_channels,
                in_query_channels * 2,
                kernel_size=3,
                padding=1,
                bias=True,
            ),
            torch.nn.ReLU(),
            nn.Conv1d(
                in_query_channels * 2,
                in_query_channels,
                kernel_size=1,
                padding=0,
                bias=True,
            ),
            torch.nn.ReLU(),
            nn.Conv1d(
                in_query_channels,
                attn_channels,
                kernel_size=1,
                padding=0,
                bias=True,
            ),
        )

        self.init_layers()

    def init_layers(self):
        r"""Initialize the weights of the key and query layers using Xavier uniform initialization.

        The gain is calculated based on the activation function: ReLU for the first layer and linear for the rest.
        """
        torch.nn.init.xavier_uniform_(
            self.key_layer[0].weight,
            gain=torch.nn.init.calculate_gain("relu"),
        )
        torch.nn.init.xavier_uniform_(
            self.key_layer[2].weight,
            gain=torch.nn.init.calculate_gain("linear"),
        )
        torch.nn.init.xavier_uniform_(
            self.query_layer[0].weight,
            gain=torch.nn.init.calculate_gain("relu"),
        )
        torch.nn.init.xavier_uniform_(
            self.query_layer[2].weight,
            gain=torch.nn.init.calculate_gain("linear"),
        )
        torch.nn.init.xavier_uniform_(
            self.query_layer[4].weight,
            gain=torch.nn.init.calculate_gain("linear"),
        )

    def _forward_aligner(
        self,
        queries: Tensor,
        keys: Tensor,
        mask: Optional[Tensor] = None,
        attn_prior: Optional[Tensor] = None,
    ) -> Tuple[Tensor, Tensor]:
        r"""Forward pass of the aligner encoder.

        Args:
            queries (Tensor): Input queries of shape [B, C, T_de].
            keys (Tensor): Input keys of shape [B, C_emb, T_en].
            mask (Optional[Tensor], optional): Mask of shape [B, T_de]. Defaults to None.
            attn_prior (Optional[Tensor], optional): Prior attention tensor. Defaults to None.

        Returns:
            Tuple[Tensor, Tensor]: A tuple containing the soft attention mask of shape [B, 1, T_en, T_de] and
            log probabilities of shape [B, 1, T_en , T_de].
        """
        key_out = self.key_layer(keys)
        query_out = self.query_layer(queries)
        attn_factor = (query_out[:, :, :, None] - key_out[:, :, None]) ** 2
        attn_logp = -self.temperature * attn_factor.sum(1, keepdim=True)
        if attn_prior is not None:
            attn_logp = self.log_softmax(attn_logp) + torch.log(
                attn_prior[:, None] + 1e-8,
            ).permute((0, 1, 3, 2))

        if mask is not None:
            attn_logp.data.masked_fill_(~mask.bool().unsqueeze(2), -float("inf"))

        attn = self.softmax(attn_logp)
        return attn, attn_logp

    def forward(
        self,
        x: Tensor,
        y: Tensor,
        x_mask: Tensor,
        y_mask: Tensor,
        attn_priors: Tensor,
    ) -> Tuple[
        Tensor,
        Tensor,
        Tensor,
        Tensor,
    ]:
        r"""Aligner forward pass.

        1. Compute a mask to apply to the attention map.
        2. Run the alignment network.
        3. Apply MAS to compute the hard alignment map.
        4. Compute the durations from the hard alignment map.

        Args:
            x (torch.Tensor): Input sequence.
            y (torch.Tensor): Output sequence.
            x_mask (torch.Tensor): Input sequence mask.
            y_mask (torch.Tensor): Output sequence mask.
            attn_priors (torch.Tensor): Prior for the aligner network map.

        Returns:
            Tuple[torch.Tensor, torch.Tensor, torch.Tensor, torch.Tensor]:
                Durations from the hard alignment map, soft alignment potentials, log scale alignment potentials,
                hard alignment map.

        Shapes:
            - x: :math:`[B, T_en, C_en]`
            - y: :math:`[B, T_de, C_de]`
            - x_mask: :math:`[B, 1, T_en]`
            - y_mask: :math:`[B, 1, T_de]`
            - attn_priors: :math:`[B, T_de, T_en]`

            - aligner_durations: :math:`[B, T_en]`
            - aligner_soft: :math:`[B, T_de, T_en]`
            - aligner_logprob: :math:`[B, 1, T_de, T_en]`
            - aligner_mas: :math:`[B, T_de, T_en]`
        """
        # [B, 1, T_en, T_de]
        attn_mask = torch.unsqueeze(x_mask, -1) * torch.unsqueeze(y_mask, 2)

        aligner_soft, aligner_logprob = self._forward_aligner(
            y.transpose(1, 2),
            x.transpose(1, 2),
            x_mask,
            attn_priors,
        )

        aligner_mas = maximum_path(
            aligner_soft.squeeze(1).transpose(1, 2).contiguous(),
            attn_mask.squeeze(1).contiguous(),
        )
        aligner_durations = torch.sum(aligner_mas, -1).int()

        # [B, T_max2, T_max]
        aligner_soft = aligner_soft.squeeze(1)
        # [B, T_max, T_max2] -> [B, T_max2, T_max]
        aligner_mas = aligner_mas.transpose(1, 2)

        return aligner_logprob, aligner_soft, aligner_mas, aligner_durations