Hugging Face's logo

437aewuh
/
so-vits-svc-fork-kiritan

Audio-to-Audio
TensorBoard
Japanese
music
Model card Files Metrics Community
10
437aewuh commited on
Commit
e2dac9d
•
1 Parent(s): bcd4b82

feat: add ms-istft-768-44k models

Browse files
Files changed (7) hide show
  1. hifigan-256-44k-3-small/D_17400.pth +3 -0
  2. hifigan-256-44k-3-small/G_17400.pth +3 -0
  3. hifigan-256-44k-3-small/config.json +98 -0
  4. hifigan-256-44k-3-small/events.out.tfevents.1681813804.njq11tmdrk.292.0 +3 -0
  5. hifigan-256-44k-3-small/events.out.tfevents.1681820690.njq11tmdrk.517.0 +3 -0
  6. hifigan-256-44k-3-small/events.out.tfevents.1681820860.njq11tmdrk.667.0 +3 -0
  7. hifigan-256-44k-3-small/events.out.tfevents.1681870176.ntqpf493rw.255.0 +3 -0
hifigan-256-44k-3-small/D_17400.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:df62c29db03a85cf3ae3e46ab9ae1b97d3b0d261535269fce25a41cfe10c5c33
3
+ size 561098249
hifigan-256-44k-3-small/G_17400.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2d0c1a3557466503c28807a2d211219870c58c7bfdfe98dab60aa07abf777601
3
+ size 87927789
hifigan-256-44k-3-small/config.json ADDED
@@ -0,0 +1,98 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "train": {
3
+ "log_interval": 200,
4
+ "eval_interval": 200,
5
+ "seed": 1234,
6
+ "epochs": 10000,
7
+ "learning_rate": 0.0003,
8
+ "betas": [
9
+ 0.8,
10
+ 0.99
11
+ ],
12
+ "eps": 1e-09,
13
+ "batch_size": 150,
14
+ "fp16_run": true,
15
+ "bf16_run": false,
16
+ "lr_decay": 0.999875,
17
+ "segment_size": 40960,
18
+ "init_lr_ratio": 1,
19
+ "warmup_epochs": 0,
20
+ "c_mel": 45,
21
+ "c_kl": 1.0,
22
+ "use_sr": true,
23
+ "max_speclen": 512,
24
+ "port": "8001",
25
+ "keep_ckpts": 3,
26
+ "num_workers": 8,
27
+ "log_version": 0,
28
+ "ckpt_name_by_step": false,
29
+ "accumulate_grad_batches": 1
30
+ },
31
+ "data": {
32
+ "training_files": "filelists/44k/train.txt",
33
+ "validation_files": "filelists/44k/val.txt",
34
+ "max_wav_value": 32768.0,
35
+ "sampling_rate": 44100,
36
+ "filter_length": 2048,
37
+ "hop_length": 512,
38
+ "win_length": 2048,
39
+ "n_mel_channels": 80,
40
+ "mel_fmin": 0.0,
41
+ "mel_fmax": 22050
42
+ },
43
+ "model": {
44
+ "inter_channels": 32,
45
+ "hidden_channels": 32,
46
+ "filter_channels": 64,
47
+ "n_heads": 2,
48
+ "n_layers": 6,
49
+ "kernel_size": 3,
50
+ "p_dropout": 0.2,
51
+ "resblock": "2",
52
+ "resblock_kernel_sizes": [
53
+ 3,
54
+ 7,
55
+ 11
56
+ ],
57
+ "resblock_dilation_sizes": [
58
+ [
59
+ 1,
60
+ 3,
61
+ 5
62
+ ],
63
+ [
64
+ 1,
65
+ 3,
66
+ 5
67
+ ],
68
+ [
69
+ 1,
70
+ 3,
71
+ 5
72
+ ]
73
+ ],
74
+ "upsample_rates": [
75
+ 8,
76
+ 8,
77
+ 2,
78
+ 2,
79
+ 2
80
+ ],
81
+ "upsample_initial_channel": 512,
82
+ "upsample_kernel_sizes": [
83
+ 16,
84
+ 16,
85
+ 4,
86
+ 4,
87
+ 4
88
+ ],
89
+ "n_layers_q": 3,
90
+ "use_spectral_norm": false,
91
+ "gin_channels": 32,
92
+ "ssl_dim": 256,
93
+ "n_speakers": 1
94
+ },
95
+ "spk": {
96
+ "kiritan": 0
97
+ }
98
+ }
hifigan-256-44k-3-small/events.out.tfevents.1681813804.njq11tmdrk.292.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:693e1493bf97ee9faec3342c0df71afcf472089fe49b78f7afb707125c976377
3
+ size 223886260
hifigan-256-44k-3-small/events.out.tfevents.1681820690.njq11tmdrk.517.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b50891b373c9b5f2f7a4200d86bacbc87832f39eb9281a67cfdc0718e5e8ac99
3
+ size 2950955
hifigan-256-44k-3-small/events.out.tfevents.1681820860.njq11tmdrk.667.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a22ede1c7ffc506c3946fb2ff045c0c5441f192a5d9c0db548d8e2796c19baef
3
+ size 51238851
hifigan-256-44k-3-small/events.out.tfevents.1681870176.ntqpf493rw.255.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d6c76bb9530fda0afe2f5782536680cfbd7b94bce2fa1c96910012c3ea2c1a1d
3
+ size 1764