ORI-Muchim commited on
Commit
e7a4e3e
1 Parent(s): 945d94b

Upload 2 files

Browse files
pretrained_model/arona_ms_istft_vits.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3bf4ea110c533a40444bc87cb449db41fc4416e2c388b4657c9ad8a196c18ebf
3
+ size 416939547
pretrained_model/arona_ms_istft_vits_config.json ADDED
@@ -0,0 +1,64 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "train": {
3
+ "log_interval": 200,
4
+ "eval_interval": 1000,
5
+ "seed": 1234,
6
+ "epochs": 20000,
7
+ "learning_rate": 2e-4,
8
+ "betas": [0.8, 0.99],
9
+ "eps": 1e-9,
10
+ "batch_size": 64,
11
+ "fp16_run": false,
12
+ "lr_decay": 0.999875,
13
+ "segment_size": 8192,
14
+ "init_lr_ratio": 1,
15
+ "warmup_epochs": 0,
16
+ "c_mel": 45,
17
+ "c_kl": 1.0,
18
+ "fft_sizes": [384, 683, 171],
19
+ "hop_sizes": [30, 60, 10],
20
+ "win_lengths": [150, 300, 60],
21
+ "window": "hann_window"
22
+ },
23
+ "data": {
24
+ "training_files":"filelists/train.txt.cleaned",
25
+ "validation_files":"filelists/val.txt.cleaned",
26
+ "text_cleaners":["japanese_cleaners2"],
27
+ "max_wav_value": 32768.0,
28
+ "sampling_rate": 44100,
29
+ "filter_length": 1024,
30
+ "hop_length": 256,
31
+ "win_length": 1024,
32
+ "n_mel_channels": 80,
33
+ "mel_fmin": 0.0,
34
+ "mel_fmax": null,
35
+ "add_blank": true,
36
+ "n_speakers": 0,
37
+ "cleaned_text": true
38
+ },
39
+ "model": {
40
+ "ms_istft_vits": true,
41
+ "mb_istft_vits": false,
42
+ "istft_vits": false,
43
+ "subbands": 4,
44
+ "gen_istft_n_fft": 16,
45
+ "gen_istft_hop_size": 4,
46
+ "inter_channels": 192,
47
+ "hidden_channels": 192,
48
+ "filter_channels": 768,
49
+ "n_heads": 2,
50
+ "n_layers": 6,
51
+ "kernel_size": 3,
52
+ "p_dropout": 0.1,
53
+ "resblock": "1",
54
+ "resblock_kernel_sizes": [3,7,11],
55
+ "resblock_dilation_sizes": [[1,3,5], [1,3,5], [1,3,5]],
56
+ "upsample_rates": [4,4],
57
+ "upsample_initial_channel": 512,
58
+ "upsample_kernel_sizes": [16,16],
59
+ "n_layers_q": 3,
60
+ "use_spectral_norm": false,
61
+ "use_sdp": false
62
+ }
63
+ }
64
+