Spaces:
Sleeping
Sleeping
File size: 8,358 Bytes
05005db |
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 |
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
# Copyright 2019 Shigeki Karita
# Apache 2.0 (http://www.apache.org/licenses/LICENSE-2.0)
"""Subsampling layer definition."""
import torch
from vec2wav2.models.conformer.embedding import PositionalEncoding
class TooShortUttError(Exception):
"""Raised when the utt is too short for subsampling.
Args:
message (str): Message for error catch
actual_size (int): the short size that cannot pass the subsampling
limit (int): the limit size for subsampling
"""
def __init__(self, message, actual_size, limit):
"""Construct a TooShortUttError for error handler."""
super().__init__(message)
self.actual_size = actual_size
self.limit = limit
def check_short_utt(ins, size):
"""Check if the utterance is too short for subsampling."""
if isinstance(ins, Conv2dSubsampling2) and size < 3:
return True, 3
if isinstance(ins, Conv2dSubsampling) and size < 7:
return True, 7
if isinstance(ins, Conv2dSubsampling6) and size < 11:
return True, 11
if isinstance(ins, Conv2dSubsampling8) and size < 15:
return True, 15
return False, -1
class Conv2dSubsampling(torch.nn.Module):
"""Convolutional 2D subsampling (to 1/4 length).
Args:
idim (int): Input dimension.
odim (int): Output dimension.
dropout_rate (float): Dropout rate.
pos_enc (torch.nn.Module): Custom position encoding layer.
"""
def __init__(self, idim, odim, dropout_rate, pos_enc=None):
"""Construct an Conv2dSubsampling object."""
super(Conv2dSubsampling, self).__init__()
self.conv = torch.nn.Sequential(
torch.nn.Conv2d(1, odim, 3, 2),
torch.nn.ReLU(),
torch.nn.Conv2d(odim, odim, 3, 2),
torch.nn.ReLU(),
)
self.out = torch.nn.Sequential(
torch.nn.Linear(odim * (((idim - 1) // 2 - 1) // 2), odim),
pos_enc if pos_enc is not None else PositionalEncoding(odim, dropout_rate),
)
def forward(self, x, x_mask):
"""Subsample x.
Args:
x (torch.Tensor): Input tensor (#batch, time, idim).
x_mask (torch.Tensor): Input mask (#batch, 1, time).
Returns:
torch.Tensor: Subsampled tensor (#batch, time', odim),
where time' = time // 4.
torch.Tensor: Subsampled mask (#batch, 1, time'),
where time' = time // 4.
"""
x = x.unsqueeze(1) # (b, c, t, f)
x = self.conv(x)
b, c, t, f = x.size()
x = self.out(x.transpose(1, 2).contiguous().view(b, t, c * f))
if x_mask is None:
return x, None
return x, x_mask[:, :, :-2:2][:, :, :-2:2]
def __getitem__(self, key):
"""Get item.
When reset_parameters() is called, if use_scaled_pos_enc is used,
return the positioning encoding.
"""
if key != -1:
raise NotImplementedError("Support only `-1` (for `reset_parameters`).")
return self.out[key]
class Conv2dSubsampling2(torch.nn.Module):
"""Convolutional 2D subsampling (to 1/2 length).
Args:
idim (int): Input dimension.
odim (int): Output dimension.
dropout_rate (float): Dropout rate.
pos_enc (torch.nn.Module): Custom position encoding layer.
"""
def __init__(self, idim, odim, dropout_rate, pos_enc=None):
"""Construct an Conv2dSubsampling2 object."""
super(Conv2dSubsampling2, self).__init__()
self.conv = torch.nn.Sequential(
torch.nn.Conv2d(1, odim, 3, 2),
torch.nn.ReLU(),
torch.nn.Conv2d(odim, odim, 3, 1),
torch.nn.ReLU(),
)
self.out = torch.nn.Sequential(
torch.nn.Linear(odim * (((idim - 1) // 2 - 2)), odim),
pos_enc if pos_enc is not None else PositionalEncoding(odim, dropout_rate),
)
def forward(self, x, x_mask):
"""Subsample x.
Args:
x (torch.Tensor): Input tensor (#batch, time, idim).
x_mask (torch.Tensor): Input mask (#batch, 1, time).
Returns:
torch.Tensor: Subsampled tensor (#batch, time', odim),
where time' = time // 2.
torch.Tensor: Subsampled mask (#batch, 1, time'),
where time' = time // 2.
"""
x = x.unsqueeze(1) # (b, c, t, f)
x = self.conv(x)
b, c, t, f = x.size()
x = self.out(x.transpose(1, 2).contiguous().view(b, t, c * f))
if x_mask is None:
return x, None
return x, x_mask[:, :, :-2:2][:, :, :-2:1]
def __getitem__(self, key):
"""Get item.
When reset_parameters() is called, if use_scaled_pos_enc is used,
return the positioning encoding.
"""
if key != -1:
raise NotImplementedError("Support only `-1` (for `reset_parameters`).")
return self.out[key]
class Conv2dSubsampling6(torch.nn.Module):
"""Convolutional 2D subsampling (to 1/6 length).
Args:
idim (int): Input dimension.
odim (int): Output dimension.
dropout_rate (float): Dropout rate.
pos_enc (torch.nn.Module): Custom position encoding layer.
"""
def __init__(self, idim, odim, dropout_rate, pos_enc=None):
"""Construct an Conv2dSubsampling6 object."""
super(Conv2dSubsampling6, self).__init__()
self.conv = torch.nn.Sequential(
torch.nn.Conv2d(1, odim, 3, 2),
torch.nn.ReLU(),
torch.nn.Conv2d(odim, odim, 5, 3),
torch.nn.ReLU(),
)
self.out = torch.nn.Sequential(
torch.nn.Linear(odim * (((idim - 1) // 2 - 2) // 3), odim),
pos_enc if pos_enc is not None else PositionalEncoding(odim, dropout_rate),
)
def forward(self, x, x_mask):
"""Subsample x.
Args:
x (torch.Tensor): Input tensor (#batch, time, idim).
x_mask (torch.Tensor): Input mask (#batch, 1, time).
Returns:
torch.Tensor: Subsampled tensor (#batch, time', odim),
where time' = time // 6.
torch.Tensor: Subsampled mask (#batch, 1, time'),
where time' = time // 6.
"""
x = x.unsqueeze(1) # (b, c, t, f)
x = self.conv(x)
b, c, t, f = x.size()
x = self.out(x.transpose(1, 2).contiguous().view(b, t, c * f))
if x_mask is None:
return x, None
return x, x_mask[:, :, :-2:2][:, :, :-4:3]
class Conv2dSubsampling8(torch.nn.Module):
"""Convolutional 2D subsampling (to 1/8 length).
Args:
idim (int): Input dimension.
odim (int): Output dimension.
dropout_rate (float): Dropout rate.
pos_enc (torch.nn.Module): Custom position encoding layer.
"""
def __init__(self, idim, odim, dropout_rate, pos_enc=None):
"""Construct an Conv2dSubsampling8 object."""
super(Conv2dSubsampling8, self).__init__()
self.conv = torch.nn.Sequential(
torch.nn.Conv2d(1, odim, 3, 2),
torch.nn.ReLU(),
torch.nn.Conv2d(odim, odim, 3, 2),
torch.nn.ReLU(),
torch.nn.Conv2d(odim, odim, 3, 2),
torch.nn.ReLU(),
)
self.out = torch.nn.Sequential(
torch.nn.Linear(odim * ((((idim - 1) // 2 - 1) // 2 - 1) // 2), odim),
pos_enc if pos_enc is not None else PositionalEncoding(odim, dropout_rate),
)
def forward(self, x, x_mask):
"""Subsample x.
Args:
x (torch.Tensor): Input tensor (#batch, time, idim).
x_mask (torch.Tensor): Input mask (#batch, 1, time).
Returns:
torch.Tensor: Subsampled tensor (#batch, time', odim),
where time' = time // 8.
torch.Tensor: Subsampled mask (#batch, 1, time'),
where time' = time // 8.
"""
x = x.unsqueeze(1) # (b, c, t, f)
x = self.conv(x)
b, c, t, f = x.size()
x = self.out(x.transpose(1, 2).contiguous().view(b, t, c * f))
if x_mask is None:
return x, None
return x, x_mask[:, :, :-2:2][:, :, :-2:2][:, :, :-2:2]
|