JotunnBurton commited on
Commit
f97f1b9
·
verified ·
1 Parent(s): 5b40b56

Upload 2 files

Browse files
Files changed (2) hide show
  1. G_6000.pth +3 -0
  2. config (3).json +110 -0
G_6000.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:26c1e954fd3fef4f025360220b48f197eb3026113477aff8c08b0b2716ca2a8e
3
+ size 913844446
config (3).json ADDED
@@ -0,0 +1,110 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "train": {
3
+ "log_interval": 200,
4
+ "eval_interval": 200,
5
+ "seed": 42,
6
+ "epochs": 800,
7
+ "learning_rate": 0.0001,
8
+ "betas": [
9
+ 0.8,
10
+ 0.99
11
+ ],
12
+ "eps": 1e-09,
13
+ "batch_size": 6,
14
+ "bf16_run": false,
15
+ "fp16_run": false,
16
+ "lr_decay": 0.99996,
17
+ "segment_size": 16384,
18
+ "init_lr_ratio": 1,
19
+ "warmup_epochs": 0,
20
+ "c_mel": 45,
21
+ "c_kl": 1.0,
22
+ "c_commit": 100,
23
+ "skip_optimizer": true,
24
+ "freeze_ZH_bert": false,
25
+ "freeze_JP_bert": false,
26
+ "freeze_EN_bert": false,
27
+ "freeze_emo": false
28
+ },
29
+ "data": {
30
+ "training_files": "data/cartethyia/train.list",
31
+ "validation_files": "data/cartethyia/val.list",
32
+ "max_wav_value": 32768.0,
33
+ "sampling_rate": 44100,
34
+ "filter_length": 2048,
35
+ "hop_length": 512,
36
+ "win_length": 2048,
37
+ "n_mel_channels": 128,
38
+ "mel_fmin": 0.0,
39
+ "mel_fmax": null,
40
+ "add_blank": true,
41
+ "n_speakers": 1,
42
+ "cleaned_text": true,
43
+ "spk2id": {
44
+ "CARTETHYIA": 0
45
+ }
46
+ },
47
+ "model": {
48
+ "use_spk_conditioned_encoder": true,
49
+ "use_noise_scaled_mas": true,
50
+ "use_mel_posterior_encoder": false,
51
+ "use_duration_discriminator": false,
52
+ "use_wavlm_discriminator": true,
53
+ "inter_channels": 192,
54
+ "hidden_channels": 192,
55
+ "filter_channels": 768,
56
+ "n_heads": 2,
57
+ "n_layers": 6,
58
+ "kernel_size": 3,
59
+ "p_dropout": 0.1,
60
+ "resblock": "1",
61
+ "resblock_kernel_sizes": [
62
+ 3,
63
+ 7,
64
+ 11
65
+ ],
66
+ "resblock_dilation_sizes": [
67
+ [
68
+ 1,
69
+ 3,
70
+ 5
71
+ ],
72
+ [
73
+ 1,
74
+ 3,
75
+ 5
76
+ ],
77
+ [
78
+ 1,
79
+ 3,
80
+ 5
81
+ ]
82
+ ],
83
+ "upsample_rates": [
84
+ 8,
85
+ 8,
86
+ 2,
87
+ 2,
88
+ 2
89
+ ],
90
+ "upsample_initial_channel": 512,
91
+ "upsample_kernel_sizes": [
92
+ 16,
93
+ 16,
94
+ 8,
95
+ 2,
96
+ 2
97
+ ],
98
+ "n_layers_q": 3,
99
+ "use_spectral_norm": false,
100
+ "gin_channels": 512,
101
+ "slm": {
102
+ "model": "./slm/wavlm-base-plus",
103
+ "sr": 16000,
104
+ "hidden": 768,
105
+ "nlayers": 13,
106
+ "initial_channel": 64
107
+ }
108
+ },
109
+ "version": "2.4"
110
+ }