Hugging Face's logo

guetLzy
/
VITS-fast-fine-tuning

Model card Files Community
1
guetLzy commited on
Commit
37ce50e
•
1 Parent(s): 19a97a2

Upload 3 files

Browse files
Files changed (3) hide show
  1. D_latest.pth +3 -0
  2. G_latest.pth +3 -0
  3. finetune_speaker.json +145 -0
D_latest.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a7c224a0a812a9bb1f062339e4d58817d241f56e7642c9623d744c53720f2fa6
3
+ size 187026467
G_latest.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2142f1b75eaf27fc2d21d5b4a0d3ae8aec1e7606bd5a770d1b73e70efd5b5c01
3
+ size 158896461
finetune_speaker.json ADDED
@@ -0,0 +1,145 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
+ "chinese_cleaners"
27
+ ],
28
+ "max_wav_value": 32768.0,
29
+ "sampling_rate": 16000,
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": 3,
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
+ "\u523b\u6674": 0,
90
+ "\u795e\u91cc\u7eeb\u534e": 1,
91
+ "\u949f\u79bb": 2
92
+ },
93
+ "symbols": [
94
+ "_",
95
+ "\uff0c",
96
+ "\u3002",
97
+ "\uff01",
98
+ "\uff1f",
99
+ "\u2014",
100
+ "\u2026",
101
+ "\u3105",
102
+ "\u3106",
103
+ "\u3107",
104
+ "\u3108",
105
+ "\u3109",
106
+ "\u310a",
107
+ "\u310b",
108
+ "\u310c",
109
+ "\u310d",
110
+ "\u310e",
111
+ "\u310f",
112
+ "\u3110",
113
+ "\u3111",
114
+ "\u3112",
115
+ "\u3113",
116
+ "\u3114",
117
+ "\u3115",
118
+ "\u3116",
119
+ "\u3117",
120
+ "\u3118",
121
+ "\u3119",
122
+ "\u311a",
123
+ "\u311b",
124
+ "\u311c",
125
+ "\u311d",
126
+ "\u311e",
127
+ "\u311f",
128
+ "\u3120",
129
+ "\u3121",
130
+ "\u3122",
131
+ "\u3123",
132
+ "\u3124",
133
+ "\u3125",
134
+ "\u3126",
135
+ "\u3127",
136
+ "\u3128",
137
+ "\u3129",
138
+ "\u02c9",
139
+ "\u02ca",
140
+ "\u02c7",
141
+ "\u02cb",
142
+ "\u02d9",
143
+ " "
144
+ ]
145
+ }