Hugging Face's logo

Spaces:
zym1
/
VITS-Umamusume-voice-synthesizer
Runtime error

App Files Community
zym1 commited on
Commit
bdd6531
•
1 Parent(s): 417bb6e

Upload finetune_speaker.json

Browse files
Files changed (1) hide show
  1. configs/finetune_speaker.json +169 -0
configs/finetune_speaker.json ADDED
@@ -0,0 +1,169 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "train": {
3
+ "log_interval": 10,
4
+ "eval_interval": 100,
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": 16,
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": "final_annotation_train.txt",
24
+ "validation_files": "final_annotation_val.txt",
25
+ "text_cleaners": [
26
+ "cjke_cleaners2"
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": 9,
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
+ "symbols": [
89
+ "_",
90
+ ",",
91
+ ".",
92
+ "!",
93
+ "?",
94
+ "-",
95
+ "~",
96
+ "\u2026",
97
+ "N",
98
+ "Q",
99
+ "a",
100
+ "b",
101
+ "d",
102
+ "e",
103
+ "f",
104
+ "g",
105
+ "h",
106
+ "i",
107
+ "j",
108
+ "k",
109
+ "l",
110
+ "m",
111
+ "n",
112
+ "o",
113
+ "p",
114
+ "s",
115
+ "t",
116
+ "u",
117
+ "v",
118
+ "w",
119
+ "x",
120
+ "y",
121
+ "z",
122
+ "\u0251",
123
+ "\u00e6",
124
+ "\u0283",
125
+ "\u0291",
126
+ "\u00e7",
127
+ "\u026f",
128
+ "\u026a",
129
+ "\u0254",
130
+ "\u025b",
131
+ "\u0279",
132
+ "\u00f0",
133
+ "\u0259",
134
+ "\u026b",
135
+ "\u0265",
136
+ "\u0278",
137
+ "\u028a",
138
+ "\u027e",
139
+ "\u0292",
140
+ "\u03b8",
141
+ "\u03b2",
142
+ "\u014b",
143
+ "\u0266",
144
+ "\u207c",
145
+ "\u02b0",
146
+ "`",
147
+ "^",
148
+ "#",
149
+ "*",
150
+ "=",
151
+ "\u02c8",
152
+ "\u02cc",
153
+ "\u2192",
154
+ "\u2193",
155
+ "\u2191",
156
+ " "
157
+ ],
158
+ "speakers": {
159
+ "\u534e\u76db\u987f": 0,
160
+ "\u5e73\u6d77": 1,
161
+ "\u8d64\u57ce": 2,
162
+ "\u978d\u5c71": 3,
163
+ "\u957f\u6625": 4,
164
+ "\u80e1\u5fb7": 5,
165
+ "\u9038\u4ed9": 6,
166
+ "\u5217\u514b\u661f\u6566": 7,
167
+ "\u6d4e\u5357": 8
168
+ }
169
+ }