Hugging Face's logo

Spaces:
Artificial-superintelligence
/
Ilaria_RVC
Paused

App Files Community
TheStinger commited on
Commit
c922820
·
verified ·
1 Parent(s): 82d966b

Upload 3 files

Browse files
Files changed (3) hide show
  1. configs/32k.json +46 -0
  2. configs/40k.json +46 -0
  3. configs/48k.json +46 -0
configs/32k.json ADDED
@@ -0,0 +1,46 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "train": {
3
+ "log_interval": 50,
4
+ "seed": 1234,
5
+ "epochs": 20000,
6
+ "learning_rate": 1e-4,
7
+ "betas": [0.8, 0.99],
8
+ "eps": 1e-9,
9
+ "batch_size": 4,
10
+ "fp16_run": true,
11
+ "lr_decay": 0.99975,
12
+ "segment_size": 12800,
13
+ "init_lr_ratio": 1,
14
+ "warmup_epochs": 100,
15
+ "c_mel": 45,
16
+ "c_kl": 1.0
17
+ },
18
+ "data": {
19
+ "max_wav_value": 32768.0,
20
+ "sampling_rate": 32000,
21
+ "filter_length": 1024,
22
+ "hop_length": 320,
23
+ "win_length": 1024,
24
+ "n_mel_channels": 80,
25
+ "mel_fmin": 0.0,
26
+ "mel_fmax": null
27
+ },
28
+ "model": {
29
+ "inter_channels": 192,
30
+ "hidden_channels": 192,
31
+ "filter_channels": 768,
32
+ "n_heads": 2,
33
+ "n_layers": 6,
34
+ "kernel_size": 3,
35
+ "p_dropout": 0,
36
+ "resblock": "1",
37
+ "resblock_kernel_sizes": [3,7,11],
38
+ "resblock_dilation_sizes": [[1,3,5], [1,3,5], [1,3,5]],
39
+ "upsample_rates": [10,8,2,2],
40
+ "upsample_initial_channel": 512,
41
+ "upsample_kernel_sizes": [20,16,4,4],
42
+ "use_spectral_norm": false,
43
+ "gin_channels": 256,
44
+ "spk_embed_dim": 109
45
+ }
46
+ }
configs/40k.json ADDED
@@ -0,0 +1,46 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "train": {
3
+ "log_interval": 50,
4
+ "seed": 1234,
5
+ "epochs": 20000,
6
+ "learning_rate": 1e-4,
7
+ "betas": [0.8, 0.99],
8
+ "eps": 1e-9,
9
+ "batch_size": 4,
10
+ "fp16_run": true,
11
+ "lr_decay": 0.99975,
12
+ "segment_size": 12800,
13
+ "init_lr_ratio": 1,
14
+ "warmup_epochs": 100,
15
+ "c_mel": 45,
16
+ "c_kl": 1.0
17
+ },
18
+ "data": {
19
+ "max_wav_value": 32768.0,
20
+ "sampling_rate": 40000,
21
+ "filter_length": 2048,
22
+ "hop_length": 400,
23
+ "win_length": 2048,
24
+ "n_mel_channels": 125,
25
+ "mel_fmin": 0.0,
26
+ "mel_fmax": null
27
+ },
28
+ "model": {
29
+ "inter_channels": 192,
30
+ "hidden_channels": 192,
31
+ "filter_channels": 768,
32
+ "n_heads": 2,
33
+ "n_layers": 6,
34
+ "kernel_size": 3,
35
+ "p_dropout": 0,
36
+ "resblock": "1",
37
+ "resblock_kernel_sizes": [3,7,11],
38
+ "resblock_dilation_sizes": [[1,3,5], [1,3,5], [1,3,5]],
39
+ "upsample_rates": [10,10,2,2],
40
+ "upsample_initial_channel": 512,
41
+ "upsample_kernel_sizes": [16,16,4,4],
42
+ "use_spectral_norm": false,
43
+ "gin_channels": 256,
44
+ "spk_embed_dim": 109
45
+ }
46
+ }
configs/48k.json ADDED
@@ -0,0 +1,46 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "train": {
3
+ "log_interval": 50,
4
+ "seed": 1234,
5
+ "epochs": 20000,
6
+ "learning_rate": 1e-4,
7
+ "betas": [0.8, 0.99],
8
+ "eps": 1e-9,
9
+ "batch_size": 4,
10
+ "fp16_run": true,
11
+ "lr_decay": 0.99975,
12
+ "segment_size": 17280,
13
+ "init_lr_ratio": 2,
14
+ "warmup_epochs": 100,
15
+ "c_mel": 45,
16
+ "c_kl": 1.0
17
+ },
18
+ "data": {
19
+ "max_wav_value": 32768.0,
20
+ "sampling_rate": 48000,
21
+ "filter_length": 2048,
22
+ "hop_length": 480,
23
+ "win_length": 2048,
24
+ "n_mel_channels": 128,
25
+ "mel_fmin": 0.0,
26
+ "mel_fmax": null
27
+ },
28
+ "model": {
29
+ "inter_channels": 192,
30
+ "hidden_channels": 192,
31
+ "filter_channels": 768,
32
+ "n_heads": 2,
33
+ "n_layers": 6,
34
+ "kernel_size": 3,
35
+ "p_dropout": 0,
36
+ "resblock": "1",
37
+ "resblock_kernel_sizes": [3,7,11],
38
+ "resblock_dilation_sizes": [[1,3,5], [1,3,5], [1,3,5]],
39
+ "upsample_rates": [12,10,2,2],
40
+ "upsample_initial_channel": 512,
41
+ "upsample_kernel_sizes": [24,20,4,4],
42
+ "use_spectral_norm": false,
43
+ "gin_channels": 256,
44
+ "spk_embed_dim": 109
45
+ }
46
+ }