Hugging Face's logo

Codec-SUPERB
/
AudioDec

Model card Files
xet
Community
AudioDec
File size: 3,456 Bytes 
63e3df4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
 
batch_length: 9600
batch_size: 16
config: config/autoencoder/symADuniv_vctk_48000_hop300.yaml
data:
  path: /mnt/home/yichiaowu/datasets/vctk_noisy/48000
  subset:
    test: clean_testset_wav
    train: clean_trainset_84spk_wav
    valid: clean_validset_84spk_wav
disable_cudnn: 'False'
discriminator_adv_loss_params:
  average_by_discriminators: false
discriminator_grad_norm: -1
discriminator_optimizer_params:
  betas:
  - 0.5
  - 0.9
  lr: 0.0002
  weight_decay: 0.0
discriminator_optimizer_type: Adam
discriminator_params:
  fft_sizes:
  - 1024
  - 2048
  - 512
  hop_sizes:
  - 120
  - 240
  - 50
  period_discriminator_params:
    bias: true
    channels: 32
    downsample_scales:
    - 3
    - 3
    - 3
    - 3
    - 1
    in_channels: 1
    kernel_sizes:
    - 5
    - 3
    max_downsample_channels: 1024
    nonlinear_activation: LeakyReLU
    nonlinear_activation_params:
      negative_slope: 0.1
    out_channels: 1
    use_spectral_norm: false
    use_weight_norm: true
  periods:
  - 2
  - 3
  - 5
  - 7
  - 11
  spectral_discriminator_params:
    bias: true
    channels: 32
    kernel_sizes:
    - - 3
      - 9
    - - 3
      - 9
    - - 3
      - 9
    - - 3
      - 9
    - - 3
      - 3
    - - 3
      - 3
    nonlinear_activation: LeakyReLU
    nonlinear_activation_params:
      negative_slope: 0.2
    strides:
    - - 1
      - 1
    - - 1
      - 2
    - - 1
      - 2
    - - 1
      - 2
    - - 1
      - 1
    - - 1
      - 1
  win_lengths:
  - 600
  - 1200
  - 240
  window: hann_window
discriminator_scheduler_params:
  gamma: 0.5
  milestones:
  - 200000
  - 400000
  - 600000
  - 800000
discriminator_scheduler_type: MultiStepLR
eval_interval_steps: 1000
exp_root: exp
feat_match_loss_params:
  average_by_discriminators: false
  average_by_layers: false
  include_final_outputs: false
generator_adv_loss_params:
  average_by_discriminators: false
generator_grad_norm: -1
generator_optimizer_params:
  betas:
  - 0.5
  - 0.9
  lr: 0.0001
  weight_decay: 0.0
generator_optimizer_type: Adam
generator_params:
  bias: true
  code_dim: 64
  codebook_num: 8
  codebook_size: 1024
  codec: audiodec
  dec_ratios:
  - 16
  - 8
  - 4
  - 2
  dec_strides:
  - 5
  - 5
  - 4
  - 3
  decode_channels: 32
  enc_ratios:
  - 2
  - 4
  - 8
  - 16
  enc_strides:
  - 3
  - 4
  - 5
  - 5
  encode_channels: 32
  input_channels: 1
  mode: causal
  output_channels: 1
  projector: conv1d
  quantier: residual_vq
generator_scheduler_params:
  gamma: 1.0
  step_size: 200000
generator_scheduler_type: StepLR
lambda_adv: 1.0
lambda_feat_match: 2.0
lambda_mel_loss: 45.0
lambda_shape_loss: 45.0
lambda_stft_loss: 45.0
lambda_vq_loss: 1.0
log_interval_steps: 100
mel_loss_params:
  fft_size: 2048
  fmax: 24000
  fmin: 0
  fs: 48000
  hop_size: 300
  log_base: null
  num_mels: 80
  win_length: null
  window: hann
model_type: symAudioDecUniv
num_workers: 2
outdir: exp/autoencoder/symADuniv_vctk_48000_hop300
paradigm: efficient
pin_memory: true
resume: ''
sampling_rate: 48000
save_interval_steps: 10000
seed: 1337
shape_loss_params:
  winlen:
  - 300
start_steps:
  discriminator: 500000
  generator: 0
stft_loss_params:
  fft_sizes:
  - 1024
  - 2048
  - 512
  hop_sizes:
  - 120
  - 240
  - 50
  win_lengths:
  - 600
  - 1200
  - 240
  window: hann_window
tag: autoencoder/symADuniv_vctk_48000_hop300
train_max_steps: 1000000
train_mode: autoencoder
use_feat_match_loss: true
use_mel_loss: true
use_shape_loss: false
use_stft_loss: false