Hugging Face's logo

AcappellaSet
/
ACappellaSet-Models

audio
music-source-separation
source-separation
htdemucs
acappella
Model card Files
xet
 Community
ACappellaSet-Models
File size: 2,480 Bytes
8617847
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
checkpoint: exp/redimnet2/expB6_v2/-LM/models/model_0.pt
data_type: shard
dataloader_args:
  batch_size: 16
  drop_last: true
  num_workers: 4
  pin_memory: false
  prefetch_factor: 4
dataset_args:
  aug_prob: 0.0
  cmvn: false
  filter: false
  frontend: tfmel
  num_frms: 600
  resample_rate: 16000
  sample_num_per_epoch: 0
  shuffle: true
  shuffle_args:
    shuffle_size: 2500
  spec_aug: false
  speed_perturb: false
  tfmel_args:
    do_preemph: true
    do_spec_aug: false
    eps: 1.0e-08
    f_max: 7600
    f_min: 20
    hop_length: 160
    n_fft: 512
    n_mels: 72
    norm_signal: true
    win_length: 400
enable_amp: true
exp_dir: exp/redimnet2/expB6_v2/-LM
gpus:
- 2
- 3
grad_accum_steps: 8
log_batch_interval: 100
loss: CrossEntropyLoss
loss_args: {}
margin_scheduler: MarginScheduler
margin_update:
  epoch_iter: 34125
  final_margin: 0.3
  fix_start_epoch: 0
  increase_start_epoch: 0
  increase_type: exp
  initial_margin: 0.3
  update_margin: false
model: ReDimNet2Wrap
model_args:
  C: 64
  F: 72
  block_1d_type: conv+att
  block_2d_type: basic_resnet
  causal: none
  compress_tconvs: true
  emb_bn: false
  embed_dim: 192
  feat_dim: 72
  fm_weigthing_type: NC
  global_context_att: true
  group_divisor: 1
  hop_length: 160
  out_channels: 224
  pooling_func: ASTP
  return_2d_output: true
  spec: null
  spec_in_channels: 1
  stages_setup:
  - - - 1
      - 1
    - 3
    - 3
    - - - 3
        - 3
    - 64
  - - - 2
      - 1
    - 4
    - 2
    - - - 3
        - 3
    - 64
  - - - 1
      - 2
    - 5
    - 2
    - - - 3
        - 3
    - 48
  - - - 2
      - 1
    - 5
    - 1
    - - - 3
        - 3
    - 48
  - - - 1
      - 2
    - 4
    - 0.75
    - - - 3
        - 3
    - 32
  - - - 2
      - 1
    - 3
    - 0.5
    - - - 3
        - 3
    - 24
model_init: null
noise_data: data/voxceleb/musan/lmdb
num_avg: 1
num_epochs: 5
optimizer: SGD
optimizer_args:
  lr: 0.0001
  momentum: 0.9
  nesterov: true
  weight_decay: 2.0e-05
projection_args:
  do_lm: false
  easy_margin: false
  embed_dim: 192
  lanbuda: 0.7
  margin_type: C
  num_class: 5994
  project_type: sphereface2
  scale: 32.0
  t: 3
reverb_data: data/voxceleb/rirs/lmdb
save_epoch_interval: 1
scheduler: ExponentialDecrease
scheduler_args:
  epoch_iter: 34125
  final_lr: 1.0e-05
  initial_lr: 0.0001
  num_epochs: 5
  scale_ratio: 4.0
  warm_from_zero: false
  warm_up_epoch: 0
seed: 42
train_data: data/voxceleb/vox2_dev/shard.list
train_label: data/voxceleb/vox2_dev/utt2spk