torchaudio 0.1.0 → 0.2.1
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/CHANGELOG.md +21 -0
- data/LICENSE.txt +1 -1
- data/README.md +63 -0
- data/ext/torchaudio/ext.cpp +26 -26
- data/ext/torchaudio/extconf.rb +1 -4
- data/lib/torchaudio.rb +75 -0
- data/lib/torchaudio/functional.rb +324 -0
- data/lib/torchaudio/transforms/amplitude_to_db.rb +27 -0
- data/lib/torchaudio/transforms/compute_deltas.rb +15 -0
- data/lib/torchaudio/transforms/fade.rb +74 -0
- data/lib/torchaudio/transforms/mel_scale.rb +39 -0
- data/lib/torchaudio/transforms/mel_spectrogram.rb +37 -0
- data/lib/torchaudio/transforms/mfcc.rb +43 -0
- data/lib/torchaudio/transforms/mu_law_decoding.rb +14 -0
- data/lib/torchaudio/transforms/mu_law_encoding.rb +14 -0
- data/lib/torchaudio/transforms/spectrogram.rb +27 -0
- data/lib/torchaudio/transforms/vol.rb +31 -0
- data/lib/torchaudio/version.rb +1 -1
- metadata +24 -69
checksums.yaml
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
---
|
2
2
|
SHA256:
|
3
|
-
metadata.gz:
|
4
|
-
data.tar.gz:
|
3
|
+
metadata.gz: 9ed4c14921f1eee18f5e08ddabfae51e09a9b5a7ef408f1dd67fdf7bfe9622fe
|
4
|
+
data.tar.gz: 1e37d5b9abed9cab7bf56a8c30a769bc8ff8f8a3e15e78bbb772847c444571b2
|
5
5
|
SHA512:
|
6
|
-
metadata.gz:
|
7
|
-
data.tar.gz:
|
6
|
+
metadata.gz: 9ca5436d7e4309dd9659fdce7ee893b122e9da96e9f7b15bf00de5dea32c635e2828a99939f04ef5bf0d9494ab89957829a65002dc3e855fa8a66f54abbbd181
|
7
|
+
data.tar.gz: d62b2a137c19d3b24facb11eda5c1b81be5841120b505877b8617bee2b9f183dbe4b4d42a95af27447346a3d48476d7faec48b57cd89b88c0ddc9709f1b5d51b
|
data/CHANGELOG.md
CHANGED
@@ -1,3 +1,24 @@
|
|
1
|
+
## 0.2.1 (2021-07-16)
|
2
|
+
|
3
|
+
- Added `create_dct` method
|
4
|
+
- Added `ComputeDeltas`, `Fade`, `MFCC`, and `Vol` transforms
|
5
|
+
|
6
|
+
## 0.2.0 (2021-05-23)
|
7
|
+
|
8
|
+
- Updated to Rice 4
|
9
|
+
- Dropped support for Ruby < 2.6
|
10
|
+
|
11
|
+
## 0.1.2 (2021-02-06)
|
12
|
+
|
13
|
+
- Added `amplitude_to_DB` and `DB_to_amplitude` methods
|
14
|
+
- Added `AmplitudeToDB` transform
|
15
|
+
- Fixed `save` options
|
16
|
+
|
17
|
+
## 0.1.1 (2020-08-26)
|
18
|
+
|
19
|
+
- Added `save` method
|
20
|
+
- Added transforms
|
21
|
+
|
1
22
|
## 0.1.0 (2020-08-24)
|
2
23
|
|
3
24
|
- First release
|
data/LICENSE.txt
CHANGED
data/README.md
CHANGED
@@ -2,6 +2,8 @@
|
|
2
2
|
|
3
3
|
:fire: An audio library for Torch.rb
|
4
4
|
|
5
|
+
[![Build Status](https://github.com/ankane/torchaudio/workflows/build/badge.svg?branch=master)](https://github.com/ankane/torchaudio/actions)
|
6
|
+
|
5
7
|
## Installation
|
6
8
|
|
7
9
|
First, [install SoX](#sox-installation). For Homebrew, use:
|
@@ -20,6 +22,67 @@ gem 'torchaudio'
|
|
20
22
|
|
21
23
|
This library follows the [Python API](https://pytorch.org/audio/). Many methods and options are missing at the moment. PRs welcome!
|
22
24
|
|
25
|
+
## Tutorial
|
26
|
+
|
27
|
+
- [PyTorch tutorial](https://pytorch.org/tutorials/beginner/audio_preprocessing_tutorial.html)
|
28
|
+
- [Ruby code](examples/tutorial.rb)
|
29
|
+
|
30
|
+
Download the [audio file](https://github.com/pytorch/tutorials/raw/master/_static/img/steam-train-whistle-daniel_simon-converted-from-mp3.wav) and install the [matplotlib](https://github.com/mrkn/matplotlib.rb) gem first.
|
31
|
+
|
32
|
+
## Basics
|
33
|
+
|
34
|
+
Load a file
|
35
|
+
|
36
|
+
```ruby
|
37
|
+
waveform, sample_rate = TorchAudio.load("file.wav")
|
38
|
+
```
|
39
|
+
|
40
|
+
Save a file
|
41
|
+
|
42
|
+
```ruby
|
43
|
+
TorchAudio.save("new.wave", waveform, sample_rate)
|
44
|
+
```
|
45
|
+
|
46
|
+
## Transforms
|
47
|
+
|
48
|
+
```ruby
|
49
|
+
TorchAudio::Transforms::Spectrogram.new.call(waveform)
|
50
|
+
```
|
51
|
+
|
52
|
+
Supported transforms are:
|
53
|
+
|
54
|
+
- AmplitudeToDB
|
55
|
+
- ComputeDeltas
|
56
|
+
- Fade
|
57
|
+
- MelScale
|
58
|
+
- MelSpectrogram
|
59
|
+
- MFCC
|
60
|
+
- MuLawDecoding
|
61
|
+
- MuLawEncoding
|
62
|
+
- Spectrogram
|
63
|
+
- Vol
|
64
|
+
|
65
|
+
## Functional
|
66
|
+
|
67
|
+
```ruby
|
68
|
+
TorchAudio::Functional.lowpass_biquad(waveform, sample_rate, cutoff_freq)
|
69
|
+
```
|
70
|
+
|
71
|
+
Supported functions are:
|
72
|
+
|
73
|
+
- amplitude_to_DB
|
74
|
+
- compute_deltas
|
75
|
+
- create_dct
|
76
|
+
- create_fb_matrix
|
77
|
+
- DB_to_amplitude
|
78
|
+
- dither
|
79
|
+
- gain
|
80
|
+
- highpass_biquad
|
81
|
+
- lowpass_biquad
|
82
|
+
- mu_law_decoding
|
83
|
+
- mu_law_encoding
|
84
|
+
- spectrogram
|
85
|
+
|
23
86
|
## Datasets
|
24
87
|
|
25
88
|
Load a dataset
|
data/ext/torchaudio/ext.cpp
CHANGED
@@ -1,33 +1,33 @@
|
|
1
1
|
#include <torchaudio/csrc/sox.h>
|
2
2
|
|
3
|
-
#include <rice/
|
4
|
-
|
5
|
-
using namespace Rice;
|
6
|
-
|
7
|
-
template<>
|
8
|
-
inline
|
9
|
-
sox_signalinfo_t* from_ruby<sox_signalinfo_t*>(Object x)
|
10
|
-
{
|
11
|
-
if (x.is_nil()) {
|
12
|
-
return nullptr;
|
13
|
-
}
|
14
|
-
throw std::runtime_error("Unsupported signalinfo");
|
15
|
-
}
|
16
|
-
|
17
|
-
template<>
|
18
|
-
inline
|
19
|
-
sox_encodinginfo_t* from_ruby<sox_encodinginfo_t*>(Object x)
|
20
|
-
{
|
21
|
-
if (x.is_nil()) {
|
22
|
-
return nullptr;
|
23
|
-
}
|
24
|
-
throw std::runtime_error("Unsupported encodinginfo");
|
25
|
-
}
|
3
|
+
#include <rice/rice.hpp>
|
4
|
+
#include <rice/stl.hpp>
|
26
5
|
|
27
6
|
extern "C"
|
28
7
|
void Init_ext()
|
29
8
|
{
|
30
|
-
|
31
|
-
|
32
|
-
|
9
|
+
auto rb_mTorchAudio = Rice::define_module("TorchAudio");
|
10
|
+
|
11
|
+
auto rb_mExt = Rice::define_module_under(rb_mTorchAudio, "Ext")
|
12
|
+
.define_singleton_function(
|
13
|
+
"read_audio_file",
|
14
|
+
[](const std::string& file_name, at::Tensor output, bool ch_first, int64_t nframes, int64_t offset, sox_signalinfo_t* si, sox_encodinginfo_t* ei, const char* ft) {
|
15
|
+
return torch::audio::read_audio_file(file_name, output, ch_first, nframes, offset, si, ei, ft);
|
16
|
+
})
|
17
|
+
.define_singleton_function(
|
18
|
+
"write_audio_file",
|
19
|
+
[](const std::string& file_name, const at::Tensor& tensor, sox_signalinfo_t* si, sox_encodinginfo_t* ei, const char* file_type) {
|
20
|
+
return torch::audio::write_audio_file(file_name, tensor, si, ei, file_type);
|
21
|
+
});
|
22
|
+
|
23
|
+
auto rb_cSignalInfo = Rice::define_class_under<sox_signalinfo_t>(rb_mExt, "SignalInfo")
|
24
|
+
.define_constructor(Rice::Constructor<sox_signalinfo_t>())
|
25
|
+
.define_method("rate", [](sox_signalinfo_t& self) { return self.rate; })
|
26
|
+
.define_method("channels", [](sox_signalinfo_t& self) { return self.channels; })
|
27
|
+
.define_method("precision", [](sox_signalinfo_t& self) { return self.precision; })
|
28
|
+
.define_method("length", [](sox_signalinfo_t& self) { return self.length; })
|
29
|
+
.define_method("rate=", [](sox_signalinfo_t& self, sox_rate_t rate) { self.rate = rate; })
|
30
|
+
.define_method("channels=", [](sox_signalinfo_t& self, unsigned channels) { self.channels = channels; })
|
31
|
+
.define_method("precision=", [](sox_signalinfo_t& self, unsigned precision) { self.precision = precision; })
|
32
|
+
.define_method("length=", [](sox_signalinfo_t& self, sox_uint64_t length) { self.length = length; });
|
33
33
|
}
|
data/ext/torchaudio/extconf.rb
CHANGED
@@ -1,8 +1,6 @@
|
|
1
1
|
require "mkmf-rice"
|
2
2
|
|
3
|
-
|
4
|
-
|
5
|
-
$CXXFLAGS += " -std=c++14"
|
3
|
+
$CXXFLAGS += " -std=c++17 $(optflags)"
|
6
4
|
|
7
5
|
abort "SoX not found" unless have_library("sox")
|
8
6
|
|
@@ -24,7 +22,6 @@ apple_clang = RbConfig::CONFIG["CC_VERSION_MESSAGE"] =~ /apple clang/i
|
|
24
22
|
|
25
23
|
# check omp first
|
26
24
|
if have_library("omp") || have_library("gomp")
|
27
|
-
$CXXFLAGS += " -DAT_PARALLEL_OPENMP=1"
|
28
25
|
$CXXFLAGS += " -Xclang" if apple_clang
|
29
26
|
$CXXFLAGS += " -fopenmp"
|
30
27
|
end
|
data/lib/torchaudio.rb
CHANGED
@@ -14,6 +14,17 @@ require "set"
|
|
14
14
|
# modules
|
15
15
|
require "torchaudio/datasets/utils"
|
16
16
|
require "torchaudio/datasets/yesno"
|
17
|
+
require "torchaudio/functional"
|
18
|
+
require "torchaudio/transforms/compute_deltas"
|
19
|
+
require "torchaudio/transforms/fade"
|
20
|
+
require "torchaudio/transforms/mel_scale"
|
21
|
+
require "torchaudio/transforms/mel_spectrogram"
|
22
|
+
require "torchaudio/transforms/mu_law_encoding"
|
23
|
+
require "torchaudio/transforms/mu_law_decoding"
|
24
|
+
require "torchaudio/transforms/spectrogram"
|
25
|
+
require "torchaudio/transforms/amplitude_to_db"
|
26
|
+
require "torchaudio/transforms/mfcc"
|
27
|
+
require "torchaudio/transforms/vol"
|
17
28
|
require "torchaudio/version"
|
18
29
|
|
19
30
|
module TorchAudio
|
@@ -73,6 +84,70 @@ module TorchAudio
|
|
73
84
|
load(filepath, **kwargs)
|
74
85
|
end
|
75
86
|
|
87
|
+
def save(filepath, src, sample_rate, precision: 16, channels_first: true)
|
88
|
+
si = Ext::SignalInfo.new
|
89
|
+
ch_idx = channels_first ? 0 : 1
|
90
|
+
si.rate = sample_rate
|
91
|
+
si.channels = src.dim == 1 ? 1 : src.size(ch_idx)
|
92
|
+
si.length = src.numel
|
93
|
+
si.precision = precision
|
94
|
+
save_encinfo(filepath, src, channels_first: channels_first, signalinfo: si)
|
95
|
+
end
|
96
|
+
|
97
|
+
def save_encinfo(filepath, src, channels_first: true, signalinfo: nil, encodinginfo: nil, filetype: nil)
|
98
|
+
ch_idx, len_idx = channels_first ? [0, 1] : [1, 0]
|
99
|
+
|
100
|
+
# check if save directory exists
|
101
|
+
abs_dirpath = File.dirname(File.expand_path(filepath))
|
102
|
+
unless Dir.exist?(abs_dirpath)
|
103
|
+
raise "Directory does not exist: #{abs_dirpath}"
|
104
|
+
end
|
105
|
+
# check that src is a CPU tensor
|
106
|
+
check_input(src)
|
107
|
+
# Check/Fix shape of source data
|
108
|
+
if src.dim == 1
|
109
|
+
# 1d tensors as assumed to be mono signals
|
110
|
+
src.unsqueeze!(ch_idx)
|
111
|
+
elsif src.dim > 2 || src.size(ch_idx) > 16
|
112
|
+
# assumes num_channels < 16
|
113
|
+
raise ArgumentError, "Expected format where C < 16, but found #{src.size}"
|
114
|
+
end
|
115
|
+
# sox stores the sample rate as a float, though practically sample rates are almost always integers
|
116
|
+
# convert integers to floats
|
117
|
+
if signalinfo
|
118
|
+
if signalinfo.rate && !signalinfo.rate.is_a?(Float)
|
119
|
+
if signalinfo.rate.to_f == signalinfo.rate
|
120
|
+
signalinfo.rate = signalinfo.rate.to_f
|
121
|
+
else
|
122
|
+
raise ArgumentError, "Sample rate should be a float or int"
|
123
|
+
end
|
124
|
+
end
|
125
|
+
# check if the bit precision (i.e. bits per sample) is an integer
|
126
|
+
if signalinfo.precision && ! signalinfo.precision.is_a?(Integer)
|
127
|
+
if signalinfo.precision.to_i == signalinfo.precision
|
128
|
+
signalinfo.precision = signalinfo.precision.to_i
|
129
|
+
else
|
130
|
+
raise ArgumentError, "Bit precision should be an integer"
|
131
|
+
end
|
132
|
+
end
|
133
|
+
end
|
134
|
+
# programs such as librosa normalize the signal, unnormalize if detected
|
135
|
+
if src.min >= -1.0 && src.max <= 1.0
|
136
|
+
src = src * (1 << 31)
|
137
|
+
src = src.long
|
138
|
+
end
|
139
|
+
# set filetype and allow for files with no extensions
|
140
|
+
extension = File.extname(filepath)
|
141
|
+
filetype = extension.length > 0 ? extension[1..-1] : filetype
|
142
|
+
# transpose from C x L -> L x C
|
143
|
+
if channels_first
|
144
|
+
src = src.transpose(1, 0)
|
145
|
+
end
|
146
|
+
# save data to file
|
147
|
+
src = src.contiguous
|
148
|
+
Ext.write_audio_file(filepath, src, signalinfo, encodinginfo, filetype)
|
149
|
+
end
|
150
|
+
|
76
151
|
private
|
77
152
|
|
78
153
|
def check_input(src)
|
@@ -0,0 +1,324 @@
|
|
1
|
+
module TorchAudio
|
2
|
+
module Functional
|
3
|
+
class << self
|
4
|
+
def spectrogram(waveform, pad, window, n_fft, hop_length, win_length, power, normalized)
|
5
|
+
if pad > 0
|
6
|
+
# TODO add "with torch.no_grad():" back when JIT supports it
|
7
|
+
waveform = Torch::NN::Functional.pad(waveform, [pad, pad], "constant")
|
8
|
+
end
|
9
|
+
|
10
|
+
# pack batch
|
11
|
+
shape = waveform.size
|
12
|
+
waveform = waveform.reshape(-1, shape[-1])
|
13
|
+
|
14
|
+
# default values are consistent with librosa.core.spectrum._spectrogram
|
15
|
+
spec_f =
|
16
|
+
Torch.stft(
|
17
|
+
waveform,
|
18
|
+
n_fft,
|
19
|
+
hop_length: hop_length,
|
20
|
+
win_length: win_length,
|
21
|
+
window: window,
|
22
|
+
center: true,
|
23
|
+
pad_mode: "reflect",
|
24
|
+
normalized: false,
|
25
|
+
onesided: true
|
26
|
+
)
|
27
|
+
|
28
|
+
# unpack batch
|
29
|
+
spec_f = spec_f.reshape(shape[0..-2] + spec_f.shape[-3..-1])
|
30
|
+
|
31
|
+
if normalized
|
32
|
+
spec_f.div!(window.pow(2.0).sum.sqrt)
|
33
|
+
end
|
34
|
+
if power
|
35
|
+
spec_f = complex_norm(spec_f, power: power)
|
36
|
+
end
|
37
|
+
|
38
|
+
spec_f
|
39
|
+
end
|
40
|
+
|
41
|
+
def mu_law_encoding(x, quantization_channels)
|
42
|
+
mu = quantization_channels - 1.0
|
43
|
+
if !x.floating_point?
|
44
|
+
x = x.to(dtype: :float)
|
45
|
+
end
|
46
|
+
mu = Torch.tensor(mu, dtype: x.dtype)
|
47
|
+
x_mu = Torch.sign(x) * Torch.log1p(mu * Torch.abs(x)) / Torch.log1p(mu)
|
48
|
+
x_mu = ((x_mu + 1) / 2 * mu + 0.5).to(dtype: :int64)
|
49
|
+
x_mu
|
50
|
+
end
|
51
|
+
|
52
|
+
def mu_law_decoding(x_mu, quantization_channels)
|
53
|
+
mu = quantization_channels - 1.0
|
54
|
+
if !x_mu.floating_point?
|
55
|
+
x_mu = x_mu.to(dtype: :float)
|
56
|
+
end
|
57
|
+
mu = Torch.tensor(mu, dtype: x_mu.dtype)
|
58
|
+
x = ((x_mu) / mu) * 2 - 1.0
|
59
|
+
x = Torch.sign(x) * (Torch.exp(Torch.abs(x) * Torch.log1p(mu)) - 1.0) / mu
|
60
|
+
x
|
61
|
+
end
|
62
|
+
|
63
|
+
def complex_norm(complex_tensor, power: 1.0)
|
64
|
+
complex_tensor.pow(2.0).sum(-1).pow(0.5 * power)
|
65
|
+
end
|
66
|
+
|
67
|
+
def create_fb_matrix(n_freqs, f_min, f_max, n_mels, sample_rate, norm: nil)
|
68
|
+
if norm && norm != "slaney"
|
69
|
+
raise ArgumentError, "norm must be one of None or 'slaney'"
|
70
|
+
end
|
71
|
+
|
72
|
+
# freq bins
|
73
|
+
# Equivalent filterbank construction by Librosa
|
74
|
+
all_freqs = Torch.linspace(0, sample_rate.div(2), n_freqs)
|
75
|
+
|
76
|
+
# calculate mel freq bins
|
77
|
+
# hertz to mel(f) is 2595. * math.log10(1. + (f / 700.))
|
78
|
+
m_min = 2595.0 * Math.log10(1.0 + (f_min / 700.0))
|
79
|
+
m_max = 2595.0 * Math.log10(1.0 + (f_max / 700.0))
|
80
|
+
m_pts = Torch.linspace(m_min, m_max, n_mels + 2)
|
81
|
+
# mel to hertz(mel) is 700. * (10**(mel / 2595.) - 1.)
|
82
|
+
f_pts = (Torch.pow(10, m_pts / 2595.0) - 1.0) * 700.0
|
83
|
+
# calculate the difference between each mel point and each stft freq point in hertz
|
84
|
+
f_diff = f_pts[1..-1] - f_pts[0...-1] # (n_mels + 1)
|
85
|
+
slopes = f_pts.unsqueeze(0) - all_freqs.unsqueeze(1) # (n_freqs, n_mels + 2)
|
86
|
+
# create overlapping triangles
|
87
|
+
zero = Torch.zeros(1)
|
88
|
+
down_slopes = (slopes[0..-1, 0...-2] * -1.0) / f_diff[0...-1] # (n_freqs, n_mels)
|
89
|
+
up_slopes = slopes[0..-1, 2..-1] / f_diff[1..-1] # (n_freqs, n_mels)
|
90
|
+
fb = Torch.max(zero, Torch.min(down_slopes, up_slopes))
|
91
|
+
|
92
|
+
if norm && norm == "slaney"
|
93
|
+
# Slaney-style mel is scaled to be approx constant energy per channel
|
94
|
+
enorm = 2.0 / (f_pts[2...(n_mels + 2)] - f_pts[:n_mels])
|
95
|
+
fb *= enorm.unsqueeze(0)
|
96
|
+
end
|
97
|
+
|
98
|
+
fb
|
99
|
+
end
|
100
|
+
|
101
|
+
def compute_deltas(specgram, win_length: 5, mode: "replicate")
|
102
|
+
device = specgram.device
|
103
|
+
dtype = specgram.dtype
|
104
|
+
|
105
|
+
# pack batch
|
106
|
+
shape = specgram.size
|
107
|
+
specgram = specgram.reshape(1, -1, shape[-1])
|
108
|
+
|
109
|
+
raise ArgumentError, "win_length must be >= 3" unless win_length >= 3
|
110
|
+
|
111
|
+
n = (win_length - 1).div(2)
|
112
|
+
|
113
|
+
# twice sum of integer squared
|
114
|
+
denom = n * (n + 1) * (2 * n + 1) / 3
|
115
|
+
|
116
|
+
specgram = Torch::NN::Functional.pad(specgram, [n, n], mode: mode)
|
117
|
+
|
118
|
+
kernel = Torch.arange(-n, n + 1, 1, device: device, dtype: dtype).repeat([specgram.shape[1], 1, 1])
|
119
|
+
|
120
|
+
output = Torch::NN::Functional.conv1d(specgram, kernel, groups: specgram.shape[1]) / denom
|
121
|
+
|
122
|
+
# unpack batch
|
123
|
+
output = output.reshape(shape)
|
124
|
+
end
|
125
|
+
|
126
|
+
def gain(waveform, gain_db: 1.0)
|
127
|
+
return waveform if gain_db == 0
|
128
|
+
|
129
|
+
ratio = 10 ** (gain_db / 20)
|
130
|
+
|
131
|
+
waveform * ratio
|
132
|
+
end
|
133
|
+
|
134
|
+
def dither(waveform, density_function: "TPDF", noise_shaping: false)
|
135
|
+
dithered = _apply_probability_distribution(waveform, density_function: density_function)
|
136
|
+
|
137
|
+
if noise_shaping
|
138
|
+
raise "Not implemented yet"
|
139
|
+
# _add_noise_shaping(dithered, waveform)
|
140
|
+
else
|
141
|
+
dithered
|
142
|
+
end
|
143
|
+
end
|
144
|
+
|
145
|
+
def biquad(waveform, b0, b1, b2, a0, a1, a2)
|
146
|
+
device = waveform.device
|
147
|
+
dtype = waveform.dtype
|
148
|
+
|
149
|
+
output_waveform = lfilter(
|
150
|
+
waveform,
|
151
|
+
Torch.tensor([a0, a1, a2], dtype: dtype, device: device),
|
152
|
+
Torch.tensor([b0, b1, b2], dtype: dtype, device: device)
|
153
|
+
)
|
154
|
+
output_waveform
|
155
|
+
end
|
156
|
+
|
157
|
+
def highpass_biquad(waveform, sample_rate, cutoff_freq, q: 0.707)
|
158
|
+
w0 = 2 * Math::PI * cutoff_freq / sample_rate
|
159
|
+
alpha = Math.sin(w0) / 2.0 / q
|
160
|
+
|
161
|
+
b0 = (1 + Math.cos(w0)) / 2
|
162
|
+
b1 = -1 - Math.cos(w0)
|
163
|
+
b2 = b0
|
164
|
+
a0 = 1 + alpha
|
165
|
+
a1 = -2 * Math.cos(w0)
|
166
|
+
a2 = 1 - alpha
|
167
|
+
biquad(waveform, b0, b1, b2, a0, a1, a2)
|
168
|
+
end
|
169
|
+
|
170
|
+
def lowpass_biquad(waveform, sample_rate, cutoff_freq, q: 0.707)
|
171
|
+
w0 = 2 * Math::PI * cutoff_freq / sample_rate
|
172
|
+
alpha = Math.sin(w0) / 2 / q
|
173
|
+
|
174
|
+
b0 = (1 - Math.cos(w0)) / 2
|
175
|
+
b1 = 1 - Math.cos(w0)
|
176
|
+
b2 = b0
|
177
|
+
a0 = 1 + alpha
|
178
|
+
a1 = -2 * Math.cos(w0)
|
179
|
+
a2 = 1 - alpha
|
180
|
+
biquad(waveform, b0, b1, b2, a0, a1, a2)
|
181
|
+
end
|
182
|
+
|
183
|
+
def lfilter(waveform, a_coeffs, b_coeffs, clamp: true)
|
184
|
+
# pack batch
|
185
|
+
shape = waveform.size
|
186
|
+
waveform = waveform.reshape(-1, shape[-1])
|
187
|
+
|
188
|
+
raise ArgumentError unless (a_coeffs.size(0) == b_coeffs.size(0))
|
189
|
+
raise ArgumentError unless (waveform.size.length == 2)
|
190
|
+
raise ArgumentError unless (waveform.device == a_coeffs.device)
|
191
|
+
raise ArgumentError unless (b_coeffs.device == a_coeffs.device)
|
192
|
+
|
193
|
+
device = waveform.device
|
194
|
+
dtype = waveform.dtype
|
195
|
+
n_channel, n_sample = waveform.size
|
196
|
+
n_order = a_coeffs.size(0)
|
197
|
+
n_sample_padded = n_sample + n_order - 1
|
198
|
+
raise ArgumentError unless (n_order > 0)
|
199
|
+
|
200
|
+
# Pad the input and create output
|
201
|
+
padded_waveform = Torch.zeros(n_channel, n_sample_padded, dtype: dtype, device: device)
|
202
|
+
padded_waveform[0..-1, (n_order - 1)..-1] = waveform
|
203
|
+
padded_output_waveform = Torch.zeros(n_channel, n_sample_padded, dtype: dtype, device: device)
|
204
|
+
|
205
|
+
# Set up the coefficients matrix
|
206
|
+
# Flip coefficients' order
|
207
|
+
a_coeffs_flipped = a_coeffs.flip([0])
|
208
|
+
b_coeffs_flipped = b_coeffs.flip([0])
|
209
|
+
|
210
|
+
# calculate windowed_input_signal in parallel
|
211
|
+
# create indices of original with shape (n_channel, n_order, n_sample)
|
212
|
+
window_idxs = Torch.arange(n_sample, device: device).unsqueeze(0) + Torch.arange(n_order, device: device).unsqueeze(1)
|
213
|
+
window_idxs = window_idxs.repeat([n_channel, 1, 1])
|
214
|
+
window_idxs += (Torch.arange(n_channel, device: device).unsqueeze(-1).unsqueeze(-1) * n_sample_padded)
|
215
|
+
window_idxs = window_idxs.long
|
216
|
+
# (n_order, ) matmul (n_channel, n_order, n_sample) -> (n_channel, n_sample)
|
217
|
+
input_signal_windows = Torch.matmul(b_coeffs_flipped, Torch.take(padded_waveform, window_idxs))
|
218
|
+
|
219
|
+
input_signal_windows.div!(a_coeffs[0])
|
220
|
+
a_coeffs_flipped.div!(a_coeffs[0])
|
221
|
+
input_signal_windows.t.each_with_index do |o0, i_sample|
|
222
|
+
windowed_output_signal = padded_output_waveform[0..-1, i_sample...(i_sample + n_order)]
|
223
|
+
o0.addmv!(windowed_output_signal, a_coeffs_flipped, alpha: -1)
|
224
|
+
padded_output_waveform[0..-1, i_sample + n_order - 1] = o0
|
225
|
+
end
|
226
|
+
|
227
|
+
output = padded_output_waveform[0..-1, (n_order - 1)..-1]
|
228
|
+
|
229
|
+
if clamp
|
230
|
+
output = Torch.clamp(output, -1.0, 1.0)
|
231
|
+
end
|
232
|
+
|
233
|
+
# unpack batch
|
234
|
+
output = output.reshape(shape[0...-1] + output.shape[-1..-1])
|
235
|
+
|
236
|
+
output
|
237
|
+
end
|
238
|
+
|
239
|
+
def amplitude_to_DB(amp, multiplier, amin, db_multiplier, top_db: nil)
|
240
|
+
db = Torch.log10(Torch.clamp(amp, min: amin)) * multiplier
|
241
|
+
db -= multiplier * db_multiplier
|
242
|
+
|
243
|
+
db = db.clamp(min: db.max.item - top_db) if top_db
|
244
|
+
|
245
|
+
db
|
246
|
+
end
|
247
|
+
|
248
|
+
def DB_to_amplitude(db, ref, power)
|
249
|
+
Torch.pow(Torch.pow(10.0, db * 0.1), power) * ref
|
250
|
+
end
|
251
|
+
|
252
|
+
def create_dct(n_mfcc, n_mels, norm: nil)
|
253
|
+
n = Torch.arange(n_mels.to_f)
|
254
|
+
k = Torch.arange(n_mfcc.to_f).unsqueeze!(1)
|
255
|
+
dct = Torch.cos((n + 0.5) * k * Math::PI / n_mels.to_f)
|
256
|
+
|
257
|
+
if norm.nil?
|
258
|
+
dct *= 2.0
|
259
|
+
else
|
260
|
+
raise ArgumentError, "Invalid DCT norm value" unless norm == :ortho
|
261
|
+
|
262
|
+
dct[0] *= 1.0 / Math.sqrt(2.0)
|
263
|
+
dct *= Math.sqrt(2.0 / n_mels)
|
264
|
+
end
|
265
|
+
|
266
|
+
dct.t
|
267
|
+
end
|
268
|
+
|
269
|
+
private
|
270
|
+
|
271
|
+
def _apply_probability_distribution(waveform, density_function: "TPDF")
|
272
|
+
# pack batch
|
273
|
+
shape = waveform.size
|
274
|
+
waveform = waveform.reshape(-1, shape[-1])
|
275
|
+
|
276
|
+
channel_size = waveform.size[0] - 1
|
277
|
+
time_size = waveform.size[-1] - 1
|
278
|
+
|
279
|
+
random_channel = channel_size > 0 ? Torch.randint(channel_size, [1]).item.to_i : 0
|
280
|
+
random_time = time_size > 0 ? Torch.randint(time_size, [1]).item.to_i : 0
|
281
|
+
|
282
|
+
number_of_bits = 16
|
283
|
+
up_scaling = 2 ** (number_of_bits - 1) - 2
|
284
|
+
signal_scaled = waveform * up_scaling
|
285
|
+
down_scaling = 2 ** (number_of_bits - 1)
|
286
|
+
|
287
|
+
signal_scaled_dis = waveform
|
288
|
+
if density_function == "RPDF"
|
289
|
+
rpdf = waveform[random_channel][random_time] - 0.5
|
290
|
+
|
291
|
+
signal_scaled_dis = signal_scaled + rpdf
|
292
|
+
elsif density_function == "GPDF"
|
293
|
+
# TODO Replace by distribution code once
|
294
|
+
# https://github.com/pytorch/pytorch/issues/29843 is resolved
|
295
|
+
# gaussian = torch.distributions.normal.Normal(torch.mean(waveform, -1), 1).sample()
|
296
|
+
|
297
|
+
num_rand_variables = 6
|
298
|
+
|
299
|
+
gaussian = waveform[random_channel][random_time]
|
300
|
+
(num_rand_variables * [time_size]).each do |ws|
|
301
|
+
rand_chan = Torch.randint(channel_size, [1]).item.to_i
|
302
|
+
gaussian += waveform[rand_chan][Torch.randint(ws, [1]).item.to_i]
|
303
|
+
end
|
304
|
+
|
305
|
+
signal_scaled_dis = signal_scaled + gaussian
|
306
|
+
else
|
307
|
+
# dtype needed for https://github.com/pytorch/pytorch/issues/32358
|
308
|
+
# TODO add support for dtype and device to bartlett_window
|
309
|
+
tpdf = Torch.bartlett_window(time_size + 1).to(signal_scaled.device, dtype: signal_scaled.dtype)
|
310
|
+
tpdf = tpdf.repeat([channel_size + 1, 1])
|
311
|
+
signal_scaled_dis = signal_scaled + tpdf
|
312
|
+
end
|
313
|
+
|
314
|
+
quantised_signal_scaled = Torch.round(signal_scaled_dis)
|
315
|
+
quantised_signal = quantised_signal_scaled / down_scaling
|
316
|
+
|
317
|
+
# unpack batch
|
318
|
+
quantised_signal.reshape(shape[0...-1] + quantised_signal.shape[-1..-1])
|
319
|
+
end
|
320
|
+
end
|
321
|
+
end
|
322
|
+
|
323
|
+
F = Functional
|
324
|
+
end
|
@@ -0,0 +1,27 @@
|
|
1
|
+
module TorchAudio
|
2
|
+
module Transforms
|
3
|
+
class AmplitudeToDB < Torch::NN::Module
|
4
|
+
def initialize(stype: :power, top_db: nil)
|
5
|
+
super()
|
6
|
+
|
7
|
+
@stype = stype
|
8
|
+
|
9
|
+
raise ArgumentError, 'top_db must be a positive numerical' if top_db && top_db.negative?
|
10
|
+
|
11
|
+
@top_db = top_db
|
12
|
+
@multiplier = stype == :power ? 10.0 : 20.0
|
13
|
+
@amin = 1e-10
|
14
|
+
@ref_value = 1.0
|
15
|
+
@db_multiplier = Math.log10([@amin, @ref_value].max)
|
16
|
+
end
|
17
|
+
|
18
|
+
def forward(amplitude_spectrogram)
|
19
|
+
F.amplitude_to_DB(
|
20
|
+
amplitude_spectrogram,
|
21
|
+
@multiplier, @amin, @db_multiplier,
|
22
|
+
top_db: @top_db
|
23
|
+
)
|
24
|
+
end
|
25
|
+
end
|
26
|
+
end
|
27
|
+
end
|
@@ -0,0 +1,15 @@
|
|
1
|
+
module TorchAudio
|
2
|
+
module Transforms
|
3
|
+
class ComputeDeltas < Torch::NN::Module
|
4
|
+
def initialize(win_length: 5, mode: "replicate")
|
5
|
+
super()
|
6
|
+
@win_length = win_length
|
7
|
+
@mode = mode
|
8
|
+
end
|
9
|
+
|
10
|
+
def forward(specgram)
|
11
|
+
F.compute_deltas(specgram, win_length: @win_length, mode: @mode)
|
12
|
+
end
|
13
|
+
end
|
14
|
+
end
|
15
|
+
end
|
@@ -0,0 +1,74 @@
|
|
1
|
+
module TorchAudio
|
2
|
+
module Transforms
|
3
|
+
class Fade < Torch::NN::Module
|
4
|
+
def initialize(fade_in_len: 0, fade_out_len: 0, fade_shape: "linear")
|
5
|
+
super()
|
6
|
+
@fade_in_len = fade_in_len
|
7
|
+
@fade_out_len = fade_out_len
|
8
|
+
@fade_shape = fade_shape
|
9
|
+
end
|
10
|
+
|
11
|
+
def forward(waveform)
|
12
|
+
waveform_length = waveform.size[-1]
|
13
|
+
device = waveform.device
|
14
|
+
fade_in(waveform_length).to(device) * fade_out(waveform_length).to(device) * waveform
|
15
|
+
end
|
16
|
+
|
17
|
+
private
|
18
|
+
|
19
|
+
def fade_in(waveform_length)
|
20
|
+
fade = Torch.linspace(0, 1, @fade_in_len)
|
21
|
+
ones = Torch.ones(waveform_length - @fade_in_len)
|
22
|
+
|
23
|
+
if @fade_shape == "linear"
|
24
|
+
fade = fade
|
25
|
+
end
|
26
|
+
|
27
|
+
if @fade_shape == "exponential"
|
28
|
+
fade = Torch.pow(2, (fade - 1)) * fade
|
29
|
+
end
|
30
|
+
|
31
|
+
if @fade_shape == "logarithmic"
|
32
|
+
fade = Torch.log10(0.1 + fade) + 1
|
33
|
+
end
|
34
|
+
|
35
|
+
if @fade_shape == "quarter_sine"
|
36
|
+
fade = Torch.sin(fade * Math::PI / 2)
|
37
|
+
end
|
38
|
+
|
39
|
+
if @fade_shape == "half_sine"
|
40
|
+
fade = Torch.sin(fade * Math::PI - Math::PI / 2) / 2 + 0.5
|
41
|
+
end
|
42
|
+
|
43
|
+
Torch.cat([fade, ones]).clamp!(0, 1)
|
44
|
+
end
|
45
|
+
|
46
|
+
def fade_out(waveform_length)
|
47
|
+
fade = Torch.linspace(0, 1, @fade_out_len)
|
48
|
+
ones = Torch.ones(waveform_length - @fade_out_len)
|
49
|
+
|
50
|
+
if @fade_shape == "linear"
|
51
|
+
fade = - fade + 1
|
52
|
+
end
|
53
|
+
|
54
|
+
if @fade_shape == "exponential"
|
55
|
+
fade = Torch.pow(2, - fade) * (1 - fade)
|
56
|
+
end
|
57
|
+
|
58
|
+
if @fade_shape == "logarithmic"
|
59
|
+
fade = Torch.log10(1.1 - fade) + 1
|
60
|
+
end
|
61
|
+
|
62
|
+
if @fade_shape == "quarter_sine"
|
63
|
+
fade = Torch.sin(fade * Math::PI / 2 + Math::PI / 2)
|
64
|
+
end
|
65
|
+
|
66
|
+
if @fade_shape == "half_sine"
|
67
|
+
fade = Torch.sin(fade * Math::PI + Math::PI / 2) / 2 + 0.5
|
68
|
+
end
|
69
|
+
|
70
|
+
Torch.cat([ones, fade]).clamp!(0, 1)
|
71
|
+
end
|
72
|
+
end
|
73
|
+
end
|
74
|
+
end
|
@@ -0,0 +1,39 @@
|
|
1
|
+
module TorchAudio
|
2
|
+
module Transforms
|
3
|
+
class MelScale < Torch::NN::Module
|
4
|
+
def initialize(n_mels: 128, sample_rate: 16000, f_min: 0.0, f_max: nil, n_stft: nil)
|
5
|
+
super()
|
6
|
+
@n_mels = n_mels
|
7
|
+
@sample_rate = sample_rate
|
8
|
+
@f_max = f_max || sample_rate.div(2).to_f
|
9
|
+
@f_min = f_min
|
10
|
+
|
11
|
+
raise ArgumentError, "Require f_min: %f < f_max: %f" % [f_min, @f_max] unless f_min <= @f_max
|
12
|
+
|
13
|
+
fb = n_stft.nil? ? Torch.empty(0) : F.create_fb_matrix(n_stft, @f_min, @f_max, @n_mels, @sample_rate)
|
14
|
+
register_buffer("fb", fb)
|
15
|
+
end
|
16
|
+
|
17
|
+
def forward(specgram)
|
18
|
+
shape = specgram.size
|
19
|
+
specgram = specgram.reshape(-1, shape[-2], shape[-1])
|
20
|
+
|
21
|
+
if @fb.numel == 0
|
22
|
+
tmp_fb = F.create_fb_matrix(specgram.size(1), @f_min, @f_max, @n_mels, @sample_rate)
|
23
|
+
# Attributes cannot be reassigned outside __init__ so workaround
|
24
|
+
@fb.resize!(tmp_fb.size)
|
25
|
+
@fb.copy!(tmp_fb)
|
26
|
+
end
|
27
|
+
|
28
|
+
# (channel, frequency, time).transpose(...) dot (frequency, n_mels)
|
29
|
+
# -> (channel, time, n_mels).transpose(...)
|
30
|
+
mel_specgram = Torch.matmul(specgram.transpose(1, 2), @fb).transpose(1, 2)
|
31
|
+
|
32
|
+
# unpack batch
|
33
|
+
mel_specgram = mel_specgram.reshape(shape[0...-2] + mel_specgram.shape[-2..-1])
|
34
|
+
|
35
|
+
mel_specgram
|
36
|
+
end
|
37
|
+
end
|
38
|
+
end
|
39
|
+
end
|
@@ -0,0 +1,37 @@
|
|
1
|
+
module TorchAudio
|
2
|
+
module Transforms
|
3
|
+
class MelSpectrogram < Torch::NN::Module
|
4
|
+
attr_reader :n_mels
|
5
|
+
|
6
|
+
def initialize(
|
7
|
+
sample_rate: 16000, n_fft: 400, win_length: nil, hop_length: nil, f_min: 0.0,
|
8
|
+
f_max: nil, pad: 0, n_mels: 128, window_fn: Torch.method(:hann_window),
|
9
|
+
power: 2.0, normalized: false, wkwargs: nil
|
10
|
+
)
|
11
|
+
|
12
|
+
super()
|
13
|
+
@sample_rate = sample_rate
|
14
|
+
@n_fft = n_fft
|
15
|
+
@win_length = win_length || n_fft
|
16
|
+
@hop_length = hop_length || @win_length.div(2)
|
17
|
+
@pad = pad
|
18
|
+
@power = power
|
19
|
+
@normalized = normalized
|
20
|
+
@n_mels = n_mels # number of mel frequency bins
|
21
|
+
@f_max = f_max
|
22
|
+
@f_min = f_min
|
23
|
+
@spectrogram =
|
24
|
+
Spectrogram.new(
|
25
|
+
n_fft: @n_fft, win_length: @win_length, hop_length: @hop_length, pad: @pad,
|
26
|
+
window_fn: window_fn, power: @power, normalized: @normalized, wkwargs: wkwargs
|
27
|
+
)
|
28
|
+
@mel_scale = MelScale.new(n_mels: @n_mels, sample_rate: @sample_rate, f_min: @f_min, f_max: @f_max, n_stft: @n_fft.div(2) + 1)
|
29
|
+
end
|
30
|
+
|
31
|
+
def forward(waveform)
|
32
|
+
specgram = @spectrogram.call(waveform)
|
33
|
+
@mel_scale.call(specgram)
|
34
|
+
end
|
35
|
+
end
|
36
|
+
end
|
37
|
+
end
|
@@ -0,0 +1,43 @@
|
|
1
|
+
module TorchAudio
|
2
|
+
module Transforms
|
3
|
+
class MFCC < Torch::NN::Module
|
4
|
+
|
5
|
+
SUPPORTED_DCT_TYPES = [2]
|
6
|
+
|
7
|
+
def initialize(sample_rate: 16000, n_mfcc: 40, dct_type: 2, norm: :ortho, log_mels: false, melkwargs: {})
|
8
|
+
super()
|
9
|
+
|
10
|
+
raise ArgumentError, "DCT type not supported: #{dct_type}" unless SUPPORTED_DCT_TYPES.include?(dct_type)
|
11
|
+
|
12
|
+
@sample_rate = sample_rate
|
13
|
+
@n_mfcc = n_mfcc
|
14
|
+
@dct_type = dct_type
|
15
|
+
@norm = norm
|
16
|
+
@top_db = 80.0
|
17
|
+
@amplitude_to_db = TorchAudio::Transforms::AmplitudeToDB.new(stype: :power, top_db: @top_db)
|
18
|
+
|
19
|
+
@melspectrogram = TorchAudio::Transforms::MelSpectrogram.new(sample_rate: @sample_rate, **melkwargs)
|
20
|
+
|
21
|
+
raise ArgumentError, "Cannot select more MFCC coefficients than # mel bins" if @n_mfcc > @melspectrogram.n_mels
|
22
|
+
|
23
|
+
dct_mat = F.create_dct(@n_mfcc, @melspectrogram.n_mels, norm: @norm)
|
24
|
+
register_buffer('dct_mat', dct_mat)
|
25
|
+
|
26
|
+
@log_mels = log_mels
|
27
|
+
end
|
28
|
+
|
29
|
+
def forward(waveform)
|
30
|
+
mel_specgram = @melspectrogram.(waveform)
|
31
|
+
if @log_mels
|
32
|
+
mel_specgram = Torch.log(mel_specgram + 1e-6)
|
33
|
+
else
|
34
|
+
mel_specgram = @amplitude_to_db.(mel_specgram)
|
35
|
+
end
|
36
|
+
|
37
|
+
Torch
|
38
|
+
.matmul(mel_specgram.transpose(-2, -1), @dct_mat)
|
39
|
+
.transpose(-2, -1)
|
40
|
+
end
|
41
|
+
end
|
42
|
+
end
|
43
|
+
end
|
@@ -0,0 +1,14 @@
|
|
1
|
+
module TorchAudio
|
2
|
+
module Transforms
|
3
|
+
class MuLawDecoding < Torch::NN::Module
|
4
|
+
def initialize(quantization_channels: 256)
|
5
|
+
super()
|
6
|
+
@quantization_channels = quantization_channels
|
7
|
+
end
|
8
|
+
|
9
|
+
def forward(x_mu)
|
10
|
+
F.mu_law_decoding(x_mu, @quantization_channels)
|
11
|
+
end
|
12
|
+
end
|
13
|
+
end
|
14
|
+
end
|
@@ -0,0 +1,14 @@
|
|
1
|
+
module TorchAudio
|
2
|
+
module Transforms
|
3
|
+
class MuLawEncoding < Torch::NN::Module
|
4
|
+
def initialize(quantization_channels: 256)
|
5
|
+
super()
|
6
|
+
@quantization_channels = quantization_channels
|
7
|
+
end
|
8
|
+
|
9
|
+
def forward(x)
|
10
|
+
F.mu_law_encoding(x, @quantization_channels)
|
11
|
+
end
|
12
|
+
end
|
13
|
+
end
|
14
|
+
end
|
@@ -0,0 +1,27 @@
|
|
1
|
+
module TorchAudio
|
2
|
+
module Transforms
|
3
|
+
class Spectrogram < Torch::NN::Module
|
4
|
+
def initialize(
|
5
|
+
n_fft: 400, win_length: nil, hop_length: nil, pad: 0,
|
6
|
+
window_fn: Torch.method(:hann_window), power: 2.0, normalized: false, wkwargs: nil
|
7
|
+
)
|
8
|
+
|
9
|
+
super()
|
10
|
+
@n_fft = n_fft
|
11
|
+
# number of FFT bins. the returned STFT result will have n_fft // 2 + 1
|
12
|
+
# number of frequecies due to onesided=True in torch.stft
|
13
|
+
@win_length = win_length || n_fft
|
14
|
+
@hop_length = hop_length || @win_length.div(2) # floor division
|
15
|
+
window = wkwargs.nil? ? window_fn.call(@win_length) : window_fn.call(@win_length, **wkwargs)
|
16
|
+
register_buffer("window", window)
|
17
|
+
@pad = pad
|
18
|
+
@power = power
|
19
|
+
@normalized = normalized
|
20
|
+
end
|
21
|
+
|
22
|
+
def forward(waveform)
|
23
|
+
F.spectrogram(waveform, @pad, @window, @n_fft, @hop_length, @win_length, @power, @normalized)
|
24
|
+
end
|
25
|
+
end
|
26
|
+
end
|
27
|
+
end
|
@@ -0,0 +1,31 @@
|
|
1
|
+
module TorchAudio
|
2
|
+
module Transforms
|
3
|
+
class Vol < Torch::NN::Module
|
4
|
+
def initialize(gain, gain_type: "amplitude")
|
5
|
+
super()
|
6
|
+
@gain = gain
|
7
|
+
@gain_type = gain_type
|
8
|
+
|
9
|
+
if ["amplitude", "power"].include?(gain_type) && gain < 0
|
10
|
+
raise ArgumentError, "If gain_type = amplitude or power, gain must be positive."
|
11
|
+
end
|
12
|
+
end
|
13
|
+
|
14
|
+
def forward(waveform)
|
15
|
+
if @gain_type == "amplitude"
|
16
|
+
waveform = waveform * @gain
|
17
|
+
end
|
18
|
+
|
19
|
+
if @gain_type == "db"
|
20
|
+
waveform = F.gain(waveform, @gain)
|
21
|
+
end
|
22
|
+
|
23
|
+
if @gain_type == "power"
|
24
|
+
waveform = F.gain(waveform, 10 * Math.log10(@gain))
|
25
|
+
end
|
26
|
+
|
27
|
+
Torch.clamp(waveform, -1, 1)
|
28
|
+
end
|
29
|
+
end
|
30
|
+
end
|
31
|
+
end
|
data/lib/torchaudio/version.rb
CHANGED
metadata
CHANGED
@@ -1,14 +1,14 @@
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
2
2
|
name: torchaudio
|
3
3
|
version: !ruby/object:Gem::Version
|
4
|
-
version: 0.1
|
4
|
+
version: 0.2.1
|
5
5
|
platform: ruby
|
6
6
|
authors:
|
7
7
|
- Andrew Kane
|
8
|
-
autorequire:
|
8
|
+
autorequire:
|
9
9
|
bindir: bin
|
10
10
|
cert_chain: []
|
11
|
-
date:
|
11
|
+
date: 2021-07-16 00:00:00.000000000 Z
|
12
12
|
dependencies:
|
13
13
|
- !ruby/object:Gem::Dependency
|
14
14
|
name: torch-rb
|
@@ -16,86 +16,30 @@ dependencies:
|
|
16
16
|
requirements:
|
17
17
|
- - ">="
|
18
18
|
- !ruby/object:Gem::Version
|
19
|
-
version: 0.
|
19
|
+
version: 0.7.0
|
20
20
|
type: :runtime
|
21
21
|
prerelease: false
|
22
22
|
version_requirements: !ruby/object:Gem::Requirement
|
23
23
|
requirements:
|
24
24
|
- - ">="
|
25
25
|
- !ruby/object:Gem::Version
|
26
|
-
version: 0.
|
26
|
+
version: 0.7.0
|
27
27
|
- !ruby/object:Gem::Dependency
|
28
28
|
name: rice
|
29
29
|
requirement: !ruby/object:Gem::Requirement
|
30
30
|
requirements:
|
31
31
|
- - ">="
|
32
32
|
- !ruby/object:Gem::Version
|
33
|
-
version:
|
33
|
+
version: 4.0.2
|
34
34
|
type: :runtime
|
35
35
|
prerelease: false
|
36
36
|
version_requirements: !ruby/object:Gem::Requirement
|
37
37
|
requirements:
|
38
38
|
- - ">="
|
39
39
|
- !ruby/object:Gem::Version
|
40
|
-
version:
|
41
|
-
|
42
|
-
|
43
|
-
requirement: !ruby/object:Gem::Requirement
|
44
|
-
requirements:
|
45
|
-
- - ">="
|
46
|
-
- !ruby/object:Gem::Version
|
47
|
-
version: '0'
|
48
|
-
type: :development
|
49
|
-
prerelease: false
|
50
|
-
version_requirements: !ruby/object:Gem::Requirement
|
51
|
-
requirements:
|
52
|
-
- - ">="
|
53
|
-
- !ruby/object:Gem::Version
|
54
|
-
version: '0'
|
55
|
-
- !ruby/object:Gem::Dependency
|
56
|
-
name: rake
|
57
|
-
requirement: !ruby/object:Gem::Requirement
|
58
|
-
requirements:
|
59
|
-
- - ">="
|
60
|
-
- !ruby/object:Gem::Version
|
61
|
-
version: '0'
|
62
|
-
type: :development
|
63
|
-
prerelease: false
|
64
|
-
version_requirements: !ruby/object:Gem::Requirement
|
65
|
-
requirements:
|
66
|
-
- - ">="
|
67
|
-
- !ruby/object:Gem::Version
|
68
|
-
version: '0'
|
69
|
-
- !ruby/object:Gem::Dependency
|
70
|
-
name: rake-compiler
|
71
|
-
requirement: !ruby/object:Gem::Requirement
|
72
|
-
requirements:
|
73
|
-
- - ">="
|
74
|
-
- !ruby/object:Gem::Version
|
75
|
-
version: '0'
|
76
|
-
type: :development
|
77
|
-
prerelease: false
|
78
|
-
version_requirements: !ruby/object:Gem::Requirement
|
79
|
-
requirements:
|
80
|
-
- - ">="
|
81
|
-
- !ruby/object:Gem::Version
|
82
|
-
version: '0'
|
83
|
-
- !ruby/object:Gem::Dependency
|
84
|
-
name: minitest
|
85
|
-
requirement: !ruby/object:Gem::Requirement
|
86
|
-
requirements:
|
87
|
-
- - ">="
|
88
|
-
- !ruby/object:Gem::Version
|
89
|
-
version: '5'
|
90
|
-
type: :development
|
91
|
-
prerelease: false
|
92
|
-
version_requirements: !ruby/object:Gem::Requirement
|
93
|
-
requirements:
|
94
|
-
- - ">="
|
95
|
-
- !ruby/object:Gem::Version
|
96
|
-
version: '5'
|
97
|
-
description:
|
98
|
-
email: andrew@chartkick.com
|
40
|
+
version: 4.0.2
|
41
|
+
description:
|
42
|
+
email: andrew@ankane.org
|
99
43
|
executables: []
|
100
44
|
extensions:
|
101
45
|
- ext/torchaudio/extconf.rb
|
@@ -118,12 +62,23 @@ files:
|
|
118
62
|
- lib/torchaudio.rb
|
119
63
|
- lib/torchaudio/datasets/utils.rb
|
120
64
|
- lib/torchaudio/datasets/yesno.rb
|
65
|
+
- lib/torchaudio/functional.rb
|
66
|
+
- lib/torchaudio/transforms/amplitude_to_db.rb
|
67
|
+
- lib/torchaudio/transforms/compute_deltas.rb
|
68
|
+
- lib/torchaudio/transforms/fade.rb
|
69
|
+
- lib/torchaudio/transforms/mel_scale.rb
|
70
|
+
- lib/torchaudio/transforms/mel_spectrogram.rb
|
71
|
+
- lib/torchaudio/transforms/mfcc.rb
|
72
|
+
- lib/torchaudio/transforms/mu_law_decoding.rb
|
73
|
+
- lib/torchaudio/transforms/mu_law_encoding.rb
|
74
|
+
- lib/torchaudio/transforms/spectrogram.rb
|
75
|
+
- lib/torchaudio/transforms/vol.rb
|
121
76
|
- lib/torchaudio/version.rb
|
122
77
|
homepage: https://github.com/ankane/torchaudio
|
123
78
|
licenses:
|
124
79
|
- BSD-2-Clause
|
125
80
|
metadata: {}
|
126
|
-
post_install_message:
|
81
|
+
post_install_message:
|
127
82
|
rdoc_options: []
|
128
83
|
require_paths:
|
129
84
|
- lib
|
@@ -131,15 +86,15 @@ required_ruby_version: !ruby/object:Gem::Requirement
|
|
131
86
|
requirements:
|
132
87
|
- - ">="
|
133
88
|
- !ruby/object:Gem::Version
|
134
|
-
version: '2.
|
89
|
+
version: '2.6'
|
135
90
|
required_rubygems_version: !ruby/object:Gem::Requirement
|
136
91
|
requirements:
|
137
92
|
- - ">="
|
138
93
|
- !ruby/object:Gem::Version
|
139
94
|
version: '0'
|
140
95
|
requirements: []
|
141
|
-
rubygems_version: 3.
|
142
|
-
signing_key:
|
96
|
+
rubygems_version: 3.2.22
|
97
|
+
signing_key:
|
143
98
|
specification_version: 4
|
144
99
|
summary: Data manipulation and transformation for audio signal processing
|
145
100
|
test_files: []
|