Elaina3 commited on
Commit
76e1acc
·
verified ·
1 Parent(s): 74b109a

Update Data/PRICONNE/config.json

Browse files
Files changed (1) hide show
  1. Data/PRICONNE/config.json +124 -124
Data/PRICONNE/config.json CHANGED
@@ -1,125 +1,125 @@
1
- {
2
- "model_name": "PRICONNE",
3
- "train": {
4
- "log_interval": 500,
5
- "eval_interval": 1000,
6
- "seed": 42,
7
- "epochs": 100,
8
- "learning_rate": 0.0001,
9
- "betas": [
10
- 0.8,
11
- 0.99
12
- ],
13
- "eps": 1e-09,
14
- "batch_size": 2,
15
- "bf16_run": false,
16
- "fp16_run": false,
17
- "lr_decay": 0.99996,
18
- "segment_size": 16384,
19
- "init_lr_ratio": 1,
20
- "warmup_epochs": 0,
21
- "c_mel": 45,
22
- "c_kl": 1.0,
23
- "c_commit": 100,
24
- "skip_optimizer": false,
25
- "freeze_ZH_bert": false,
26
- "freeze_JP_bert": false,
27
- "freeze_EN_bert": false,
28
- "freeze_emo": false,
29
- "freeze_style": false,
30
- "freeze_decoder": false
31
- },
32
- "data": {
33
- "use_jp_extra": true,
34
- "training_files": "Data/PRICONNE/train.list",
35
- "validation_files": "Data/PRICONNE/val.list",
36
- "max_wav_value": 32768.0,
37
- "sampling_rate": 44100,
38
- "filter_length": 2048,
39
- "hop_length": 512,
40
- "win_length": 2048,
41
- "n_mel_channels": 128,
42
- "mel_fmin": 0.0,
43
- "mel_fmax": null,
44
- "add_blank": true,
45
- "n_speakers": 12,
46
- "cleaned_text": true,
47
- "spk2id": {
48
- "アメス": 0,
49
- "イノリ": 1,
50
- "キャル": 2,
51
- "コッコロ": 3,
52
- "チエル": 4,
53
- "ハツネ": 5,
54
- "ペコリーヌ": 6,
55
- "ミヤコ": 7,
56
- "ユニ": 8,
57
- "リン": 9,
58
- "ルゥ": 10,
59
- "由仁": 11
60
- }
61
- },
62
- "model": {
63
- "use_spk_conditioned_encoder": true,
64
- "use_noise_scaled_mas": true,
65
- "use_mel_posterior_encoder": false,
66
- "use_duration_discriminator": false,
67
- "use_wavlm_discriminator": true,
68
- "inter_channels": 192,
69
- "hidden_channels": 192,
70
- "filter_channels": 768,
71
- "n_heads": 2,
72
- "n_layers": 6,
73
- "kernel_size": 3,
74
- "p_dropout": 0.1,
75
- "resblock": "1",
76
- "resblock_kernel_sizes": [
77
- 3,
78
- 7,
79
- 11
80
- ],
81
- "resblock_dilation_sizes": [
82
- [
83
- 1,
84
- 3,
85
- 5
86
- ],
87
- [
88
- 1,
89
- 3,
90
- 5
91
- ],
92
- [
93
- 1,
94
- 3,
95
- 5
96
- ]
97
- ],
98
- "upsample_rates": [
99
- 8,
100
- 8,
101
- 2,
102
- 2,
103
- 2
104
- ],
105
- "upsample_initial_channel": 512,
106
- "upsample_kernel_sizes": [
107
- 16,
108
- 16,
109
- 8,
110
- 2,
111
- 2
112
- ],
113
- "n_layers_q": 3,
114
- "use_spectral_norm": false,
115
- "gin_channels": 512,
116
- "slm": {
117
- "model": "./slm/wavlm-base-plus",
118
- "sr": 16000,
119
- "hidden": 768,
120
- "nlayers": 13,
121
- "initial_channel": 64
122
- }
123
- },
124
- "version": "2.6.1-JP-Extra"
125
  }
 
1
+ {
2
+ "model_name": "PRICONNE",
3
+ "train": {
4
+ "log_interval": 500,
5
+ "eval_interval": 1000,
6
+ "seed": 42,
7
+ "epochs": 100,
8
+ "learning_rate": 0.0001,
9
+ "betas": [
10
+ 0.8,
11
+ 0.99
12
+ ],
13
+ "eps": 1e-09,
14
+ "batch_size": 4,
15
+ "bf16_run": false,
16
+ "fp16_run": false,
17
+ "lr_decay": 0.99996,
18
+ "segment_size": 16384,
19
+ "init_lr_ratio": 1,
20
+ "warmup_epochs": 0,
21
+ "c_mel": 45,
22
+ "c_kl": 1.0,
23
+ "c_commit": 100,
24
+ "skip_optimizer": false,
25
+ "freeze_ZH_bert": false,
26
+ "freeze_JP_bert": false,
27
+ "freeze_EN_bert": false,
28
+ "freeze_emo": false,
29
+ "freeze_style": false,
30
+ "freeze_decoder": false
31
+ },
32
+ "data": {
33
+ "use_jp_extra": true,
34
+ "training_files": "Data/PRICONNE/train.list",
35
+ "validation_files": "Data/PRICONNE/val.list",
36
+ "max_wav_value": 32768.0,
37
+ "sampling_rate": 44100,
38
+ "filter_length": 2048,
39
+ "hop_length": 512,
40
+ "win_length": 2048,
41
+ "n_mel_channels": 128,
42
+ "mel_fmin": 0.0,
43
+ "mel_fmax": null,
44
+ "add_blank": true,
45
+ "n_speakers": 12,
46
+ "cleaned_text": true,
47
+ "spk2id": {
48
+ "アメス": 0,
49
+ "イノリ": 1,
50
+ "キャル": 2,
51
+ "コッコロ": 3,
52
+ "チエル": 4,
53
+ "ハツネ": 5,
54
+ "ペコリーヌ": 6,
55
+ "ミヤコ": 7,
56
+ "ユニ": 8,
57
+ "リン": 9,
58
+ "ルゥ": 10,
59
+ "由仁": 11
60
+ }
61
+ },
62
+ "model": {
63
+ "use_spk_conditioned_encoder": true,
64
+ "use_noise_scaled_mas": true,
65
+ "use_mel_posterior_encoder": false,
66
+ "use_duration_discriminator": false,
67
+ "use_wavlm_discriminator": true,
68
+ "inter_channels": 192,
69
+ "hidden_channels": 192,
70
+ "filter_channels": 768,
71
+ "n_heads": 2,
72
+ "n_layers": 6,
73
+ "kernel_size": 3,
74
+ "p_dropout": 0.1,
75
+ "resblock": "1",
76
+ "resblock_kernel_sizes": [
77
+ 3,
78
+ 7,
79
+ 11
80
+ ],
81
+ "resblock_dilation_sizes": [
82
+ [
83
+ 1,
84
+ 3,
85
+ 5
86
+ ],
87
+ [
88
+ 1,
89
+ 3,
90
+ 5
91
+ ],
92
+ [
93
+ 1,
94
+ 3,
95
+ 5
96
+ ]
97
+ ],
98
+ "upsample_rates": [
99
+ 8,
100
+ 8,
101
+ 2,
102
+ 2,
103
+ 2
104
+ ],
105
+ "upsample_initial_channel": 512,
106
+ "upsample_kernel_sizes": [
107
+ 16,
108
+ 16,
109
+ 8,
110
+ 2,
111
+ 2
112
+ ],
113
+ "n_layers_q": 3,
114
+ "use_spectral_norm": false,
115
+ "gin_channels": 512,
116
+ "slm": {
117
+ "model": "./slm/wavlm-base-plus",
118
+ "sr": 16000,
119
+ "hidden": 768,
120
+ "nlayers": 13,
121
+ "initial_channel": 64
122
+ }
123
+ },
124
+ "version": "2.6.1-JP-Extra"
125
  }