Redmonm commited on
Commit
01555a8
1 Parent(s): 4969084

Upload 2 files

Browse files
Files changed (2) hide show
  1. G_latest.pth +3 -0
  2. finetune_speaker.json +150 -0
G_latest.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f340b6731f76356df956a525d52cb96cfd41fe7f311d2ea97be42f153f9496a3
3
+ size 158902940
finetune_speaker.json ADDED
@@ -0,0 +1,150 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
+ "zh_ja_mixture_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": 6,
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
+ "\u681e\u90a3": 0,
90
+ "\u590f\u76ee": 1,
91
+ "\u5c0f\u5e0c": 2,
92
+ "\u7231\u8863": 3,
93
+ "\u51c9\u97f3": 4,
94
+ "\u5fa1\u5e1d": 5
95
+ },
96
+ "symbols": [
97
+ "_",
98
+ ",",
99
+ ".",
100
+ "!",
101
+ "?",
102
+ "-",
103
+ "~",
104
+ "\u2026",
105
+ "A",
106
+ "E",
107
+ "I",
108
+ "N",
109
+ "O",
110
+ "Q",
111
+ "U",
112
+ "a",
113
+ "b",
114
+ "d",
115
+ "e",
116
+ "f",
117
+ "g",
118
+ "h",
119
+ "i",
120
+ "j",
121
+ "k",
122
+ "l",
123
+ "m",
124
+ "n",
125
+ "o",
126
+ "p",
127
+ "r",
128
+ "s",
129
+ "t",
130
+ "u",
131
+ "v",
132
+ "w",
133
+ "y",
134
+ "z",
135
+ "\u0283",
136
+ "\u02a7",
137
+ "\u02a6",
138
+ "\u026f",
139
+ "\u0279",
140
+ "\u0259",
141
+ "\u0265",
142
+ "\u207c",
143
+ "\u02b0",
144
+ "`",
145
+ "\u2192",
146
+ "\u2193",
147
+ "\u2191",
148
+ " "
149
+ ]
150
+ }