• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1"""
2/* Copyright (c) 2023 Amazon
3   Written by Jan Buethe */
4/*
5   Redistribution and use in source and binary forms, with or without
6   modification, are permitted provided that the following conditions
7   are met:
8
9   - Redistributions of source code must retain the above copyright
10   notice, this list of conditions and the following disclaimer.
11
12   - Redistributions in binary form must reproduce the above copyright
13   notice, this list of conditions and the following disclaimer in the
14   documentation and/or other materials provided with the distribution.
15
16   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
17   ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
18   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
19   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
20   OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
21   EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
22   PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
23   PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
24   LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
25   NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
26   SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
27*/
28"""
29
30import torch
31from torch import nn
32import torch.nn.functional as F
33
34from utils.complexity import _conv1d_flop_count
35
36class NoiseShaper(nn.Module):
37
38    def __init__(self,
39                 feature_dim,
40                 frame_size=160
41    ):
42        """
43
44        Parameters:
45        -----------
46
47        feature_dim : int
48            dimension of input features
49
50        frame_size : int
51            frame size
52
53        """
54
55        super().__init__()
56
57        self.feature_dim    = feature_dim
58        self.frame_size     = frame_size
59
60        # feature transform
61        self.feature_alpha1 = nn.Conv1d(self.feature_dim, frame_size, 2)
62        self.feature_alpha2 = nn.Conv1d(frame_size, frame_size, 2)
63
64
65    def flop_count(self, rate):
66
67        frame_rate = rate / self.frame_size
68
69        shape_flops = sum([_conv1d_flop_count(x, frame_rate) for x in (self.feature_alpha1, self.feature_alpha2)]) + 11 * frame_rate * self.frame_size
70
71        return shape_flops
72
73
74    def forward(self, features):
75        """ creates temporally shaped noise
76
77
78        Parameters:
79        -----------
80        features : torch.tensor
81            frame-wise features of shape (batch_size, num_frames, feature_dim)
82
83        """
84
85        batch_size = features.size(0)
86        num_frames = features.size(1)
87        frame_size = self.frame_size
88        num_samples = num_frames * frame_size
89
90        # feature path
91        f = F.pad(features.permute(0, 2, 1), [1, 0])
92        alpha = F.leaky_relu(self.feature_alpha1(f), 0.2)
93        alpha = torch.exp(self.feature_alpha2(F.pad(alpha, [1, 0])))
94        alpha = alpha.permute(0, 2, 1)
95
96        # signal generation
97        y = torch.randn((batch_size, num_frames, frame_size), dtype=features.dtype, device=features.device)
98        y = alpha * y
99
100        return y.reshape(batch_size, 1, num_samples)
101