models.rst 3.27 KB
Newer Older
Tomás Osório's avatar
Tomás Osório committed
1
2
3
4
.. role:: hidden
    :class: hidden-section

torchaudio.models
moto's avatar
moto committed
5
=================
Tomás Osório's avatar
Tomás Osório committed
6
7
8
9
10
11

.. currentmodule:: torchaudio.models

The models subpackage contains definitions of models for addressing common audio tasks.


moto's avatar
moto committed
12
13
ConvTasNet
~~~~~~~~~~
moto's avatar
moto committed
14
15
16
17
18
19

.. autoclass:: ConvTasNet

  .. automethod:: forward


moto's avatar
moto committed
20
21
DeepSpeech
~~~~~~~~~~
discort's avatar
discort committed
22
23
24
25
26
27

.. autoclass:: DeepSpeech

  .. automethod:: forward


28
29
30
Tacotron2
~~~~~~~~~

moto's avatar
moto committed
31
32
33
34
35
36
Model
-----

Tacotoron2
^^^^^^^^^^

37
38
39
40
41
42
43
44
45
46
.. autoclass:: Tacotron2

  .. automethod:: forward

  .. automethod:: infer

Factory Functions
-----------------

tacotron2
moto's avatar
moto committed
47
^^^^^^^^^
48
49
50
51

.. autofunction:: tacotron2


moto's avatar
moto committed
52
53
Wav2Letter
~~~~~~~~~~
Tomás Osório's avatar
Tomás Osório committed
54
55
56
57

.. autoclass:: Wav2Letter

  .. automethod:: forward
jimchen90's avatar
jimchen90 committed
58
59


moto's avatar
moto committed
60
61
Wav2Vec2.0 / HuBERT
~~~~~~~~~~~~~~~~~~~
moto's avatar
moto committed
62

moto's avatar
moto committed
63
64
65
Model
-----

moto's avatar
moto committed
66
Wav2Vec2Model
moto's avatar
moto committed
67
^^^^^^^^^^^^^
moto's avatar
moto committed
68
69
70
71
72
73
74
75
76
77

.. autoclass:: Wav2Vec2Model

  .. automethod:: extract_features

  .. automethod:: forward

Factory Functions
-----------------

moto's avatar
moto committed
78
wav2vec2_base
moto's avatar
moto committed
79
^^^^^^^^^^^^^
moto's avatar
moto committed
80

moto's avatar
moto committed
81
82
.. autofunction:: wav2vec2_base

moto's avatar
moto committed
83
wav2vec2_large
moto's avatar
moto committed
84
^^^^^^^^^^^^^^
moto's avatar
moto committed
85

moto's avatar
moto committed
86
87
.. autofunction:: wav2vec2_large

moto's avatar
moto committed
88
wav2vec2_large_lv60k
moto's avatar
moto committed
89
^^^^^^^^^^^^^^^^^^^^
moto's avatar
moto committed
90

moto's avatar
moto committed
91
92
.. autofunction:: wav2vec2_large_lv60k

93
wav2vec2_ft_base
94
95
^^^^^^^^^^^^^^^^^

96
.. autofunction:: wav2vec2_ft_base
97

98
wav2vec2_ft_large
99
100
^^^^^^^^^^^^^^^^^^

101
.. autofunction:: wav2vec2_ft_large
102

103
wav2vec2_ft_large_lv60k
104
105
^^^^^^^^^^^^^^^^^^^^^^^^

106
.. autofunction:: wav2vec2_ft_large_lv60k
107

moto's avatar
moto committed
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
hubert_base
^^^^^^^^^^^

.. autofunction:: hubert_base

hubert_large
^^^^^^^^^^^^

.. autofunction:: hubert_large

hubert_xlarge
^^^^^^^^^^^^^

.. autofunction:: hubert_xlarge

123
hubert_ft_large
moto's avatar
moto committed
124
125
^^^^^^^^^^^^^^^^

126
.. autofunction:: hubert_ft_large
moto's avatar
moto committed
127

128
hubert_ft_xlarge
moto's avatar
moto committed
129
130
^^^^^^^^^^^^^^^^^

131
.. autofunction:: hubert_ft_xlarge
moto's avatar
moto committed
132

133
134
135
136
137
138
139
140
141
Pre-trained Models
------------------

.. autoclass:: Wav2Vec2PretrainedModelBundle

   .. automethod:: get_model

   .. autoproperty:: labels

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
   .. autodata:: WAV2VEC2_BASE
      :no-value:

   .. autodata:: WAV2VEC2_ASR_BASE_10M
      :no-value:

   .. autodata:: WAV2VEC2_ASR_BASE_100H
      :no-value:

   .. autodata:: WAV2VEC2_ASR_BASE_960H
      :no-value:

   .. autodata:: WAV2VEC2_LARGE
      :no-value:

   .. autodata:: WAV2VEC2_ASR_LARGE_10M
      :no-value:

   .. autodata:: WAV2VEC2_ASR_LARGE_100H
      :no-value:

   .. autodata:: WAV2VEC2_ASR_LARGE_960H
      :no-value:

   .. autodata:: WAV2VEC2_LARGE_LV60K
      :no-value:

   .. autodata:: WAV2VEC2_ASR_LARGE_LV60K_10M
      :no-value:

   .. autodata:: WAV2VEC2_ASR_LARGE_LV60K_100H
      :no-value:

   .. autodata:: WAV2VEC2_ASR_LARGE_LV60K_960H
      :no-value:

   .. autodata:: WAV2VEC2_XLSR53
      :no-value:

181
182
183
   .. autodata:: HUBERT_BASE
      :no-value:

184
185
186
187
188
189
   .. autodata:: HUBERT_LARGE
      :no-value:

   .. autodata:: HUBERT_XLARGE
      :no-value:

190
191
   .. autodata:: HUBERT_ASR_LARGE
      :no-value:
192

193
194
195
   .. autodata:: HUBERT_ASR_XLARGE
      :no-value:

196
197
198
Utility Functions
-----------------

199
200
.. currentmodule:: torchaudio.models.wav2vec2.utils

moto's avatar
moto committed
201
import_huggingface_model
moto's avatar
moto committed
202
^^^^^^^^^^^^^^^^^^^^^^^^
moto's avatar
moto committed
203

204
205
.. autofunction:: import_huggingface_model

moto's avatar
moto committed
206
import_fairseq_model
moto's avatar
moto committed
207
^^^^^^^^^^^^^^^^^^^^
moto's avatar
moto committed
208

moto's avatar
moto committed
209
210
.. autofunction:: import_fairseq_model

211
.. currentmodule:: torchaudio.models
moto's avatar
moto committed
212

moto's avatar
moto committed
213
214
WaveRNN
~~~~~~~
jimchen90's avatar
jimchen90 committed
215

moto's avatar
moto committed
216
217
218
219
220
221
Model
-----

WaveRNN
^^^^^^^

jimchen90's avatar
jimchen90 committed
222
223
224
.. autoclass:: WaveRNN

  .. automethod:: forward
moto's avatar
moto committed
225

226
227
  .. automethod:: infer

228
229
230
231
Factory Functions
-----------------

wavernn
moto's avatar
moto committed
232
^^^^^^^
233
234
235

.. autofunction:: wavernn

moto's avatar
moto committed
236
237
238
239
References
~~~~~~~~~~

.. footbibliography::