ORI-Muchim commited on
Commit
83e6204
β€’
1 Parent(s): 544d3d7

Upload 3 files

Browse files
ko_fine_tuning_22050hz/D_172000.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8fec796f8088b43a691357d2cc1d12d53c725e5c2df17efb2aae26fcfeb65612
3
+ size 561099143
ko_fine_tuning_22050hz/G_172000.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:356fdbc3390d035117275b011bd82ca14af20ebf69d49be711bcf8691d73f1c5
3
+ size 476986523
ko_fine_tuning_22050hz/config.json ADDED
@@ -0,0 +1,131 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "train": {
3
+ "log_interval": 200,
4
+ "eval_interval": 1000,
5
+ "seed": 1234,
6
+ "epochs": 10000,
7
+ "learning_rate": 0.0002,
8
+ "betas": [
9
+ 0.8,
10
+ 0.99
11
+ ],
12
+ "eps": 1e-09,
13
+ "batch_size": 32,
14
+ "fp16_run": true,
15
+ "lr_decay": 0.999875,
16
+ "segment_size": 8192,
17
+ "init_lr_ratio": 1,
18
+ "warmup_epochs": 0,
19
+ "c_mel": 45,
20
+ "c_kl": 1.0
21
+ },
22
+ "data": {
23
+ "training_files": "../datasets/kss_train.txt.cleaned",
24
+ "validation_files": "../datasets/kss_val.txt.cleaned",
25
+ "text_cleaners": [
26
+ "korean_cleaners"
27
+ ],
28
+ "max_wav_value": 32768.0,
29
+ "sampling_rate": 22050,
30
+ "filter_length": 1024,
31
+ "hop_length": 256,
32
+ "win_length": 1024,
33
+ "n_mel_channels": 80,
34
+ "mel_fmin": 0.0,
35
+ "mel_fmax": null,
36
+ "add_blank": true,
37
+ "n_speakers": 70,
38
+ "cleaned_text": true
39
+ },
40
+ "model": {
41
+ "inter_channels": 192,
42
+ "hidden_channels": 192,
43
+ "filter_channels": 768,
44
+ "n_heads": 2,
45
+ "n_layers": 6,
46
+ "kernel_size": 3,
47
+ "p_dropout": 0.1,
48
+ "resblock": "1",
49
+ "resblock_kernel_sizes": [
50
+ 3,
51
+ 7,
52
+ 11
53
+ ],
54
+ "resblock_dilation_sizes": [
55
+ [
56
+ 1,
57
+ 3,
58
+ 5
59
+ ],
60
+ [
61
+ 1,
62
+ 3,
63
+ 5
64
+ ],
65
+ [
66
+ 1,
67
+ 3,
68
+ 5
69
+ ]
70
+ ],
71
+ "upsample_rates": [
72
+ 8,
73
+ 8,
74
+ 2,
75
+ 2
76
+ ],
77
+ "upsample_initial_channel": 512,
78
+ "upsample_kernel_sizes": [
79
+ 16,
80
+ 16,
81
+ 4,
82
+ 4
83
+ ],
84
+ "n_layers_q": 3,
85
+ "use_spectral_norm": false,
86
+ "gin_channels": 256
87
+ },
88
+ "speakers": [
89
+ "1",
90
+ "2",
91
+ "4",
92
+ "3"
93
+ ],
94
+ "symbols": [
95
+ "_",
96
+ ",",
97
+ ".",
98
+ "!",
99
+ "?",
100
+ "…",
101
+ "~",
102
+ "γ„±",
103
+ "γ„΄",
104
+ "γ„·",
105
+ "γ„Ή",
106
+ "ㅁ",
107
+ "γ…‚",
108
+ "γ……",
109
+ "γ…‡",
110
+ "γ…ˆ",
111
+ "γ…Š",
112
+ "γ…‹",
113
+ "γ…Œ",
114
+ "ㅍ",
115
+ "γ…Ž",
116
+ "γ„²",
117
+ "γ„Έ",
118
+ "γ…ƒ",
119
+ "γ…†",
120
+ "γ…‰",
121
+ "ㅏ",
122
+ "γ…“",
123
+ "γ…—",
124
+ "γ…œ",
125
+ "γ…‘",
126
+ "γ…£",
127
+ "ㅐ",
128
+ "γ…”",
129
+ " "
130
+ ]
131
+ }