AbeShinzo0708 commited on
Commit
9b8323d
1 Parent(s): b8612d0

Upload 5 files

Browse files
AbeShinzo20240210_e300_s43800.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f50af9e0fb633d577046581c945bae195464de49270b4d2587f3a26b46b6ccd7
3
+ size 251150980
config.json ADDED
@@ -0,0 +1,121 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "train": {
3
+ "log_interval": 200,
4
+ "eval_interval": 1000,
5
+ "seed": 42,
6
+ "epochs": 300,
7
+ "learning_rate": 0.0001,
8
+ "betas": [
9
+ 0.8,
10
+ 0.99
11
+ ],
12
+ "eps": 1e-09,
13
+ "batch_size": 2,
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": false,
24
+ "freeze_ZH_bert": false,
25
+ "freeze_JP_bert": false,
26
+ "freeze_EN_bert": false,
27
+ "freeze_emo": false,
28
+ "freeze_style": false
29
+ },
30
+ "data": {
31
+ "use_jp_extra": true,
32
+ "training_files": "Data/AbeShinzo20240210/train.list",
33
+ "validation_files": "Data/AbeShinzo20240210/val.list",
34
+ "max_wav_value": 32768.0,
35
+ "sampling_rate": 44100,
36
+ "filter_length": 2048,
37
+ "hop_length": 512,
38
+ "win_length": 2048,
39
+ "n_mel_channels": 128,
40
+ "mel_fmin": 0.0,
41
+ "mel_fmax": null,
42
+ "add_blank": true,
43
+ "n_speakers": 1,
44
+ "cleaned_text": true,
45
+ "spk2id": {
46
+ "AbeShinzo20240210": 0
47
+ },
48
+ "num_styles": 5,
49
+ "style2id": {
50
+ "Neutral": 0,
51
+ "Angry": 1,
52
+ "Sad": 2,
53
+ "Noisy": 3,
54
+ "Calm": 4
55
+ }
56
+ },
57
+ "model": {
58
+ "use_spk_conditioned_encoder": true,
59
+ "use_noise_scaled_mas": true,
60
+ "use_mel_posterior_encoder": false,
61
+ "use_duration_discriminator": false,
62
+ "use_wavlm_discriminator": true,
63
+ "inter_channels": 192,
64
+ "hidden_channels": 192,
65
+ "filter_channels": 768,
66
+ "n_heads": 2,
67
+ "n_layers": 6,
68
+ "kernel_size": 3,
69
+ "p_dropout": 0.1,
70
+ "resblock": "1",
71
+ "resblock_kernel_sizes": [
72
+ 3,
73
+ 7,
74
+ 11
75
+ ],
76
+ "resblock_dilation_sizes": [
77
+ [
78
+ 1,
79
+ 3,
80
+ 5
81
+ ],
82
+ [
83
+ 1,
84
+ 3,
85
+ 5
86
+ ],
87
+ [
88
+ 1,
89
+ 3,
90
+ 5
91
+ ]
92
+ ],
93
+ "upsample_rates": [
94
+ 8,
95
+ 8,
96
+ 2,
97
+ 2,
98
+ 2
99
+ ],
100
+ "upsample_initial_channel": 512,
101
+ "upsample_kernel_sizes": [
102
+ 16,
103
+ 16,
104
+ 8,
105
+ 2,
106
+ 2
107
+ ],
108
+ "n_layers_q": 3,
109
+ "use_spectral_norm": false,
110
+ "gin_channels": 512,
111
+ "slm": {
112
+ "model": "./slm/wavlm-base-plus",
113
+ "sr": 16000,
114
+ "hidden": 768,
115
+ "nlayers": 13,
116
+ "initial_channel": 64
117
+ }
118
+ },
119
+ "version": "2.2-JP-Extra",
120
+ "model_name": "AbeShinzo20240210"
121
+ }
config.json.bak ADDED
@@ -0,0 +1,117 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "train": {
3
+ "log_interval": 200,
4
+ "eval_interval": 1000,
5
+ "seed": 42,
6
+ "epochs": 300,
7
+ "learning_rate": 0.0001,
8
+ "betas": [
9
+ 0.8,
10
+ 0.99
11
+ ],
12
+ "eps": 1e-09,
13
+ "batch_size": 2,
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": false,
24
+ "freeze_ZH_bert": false,
25
+ "freeze_JP_bert": false,
26
+ "freeze_EN_bert": false,
27
+ "freeze_emo": false,
28
+ "freeze_style": false
29
+ },
30
+ "data": {
31
+ "use_jp_extra": true,
32
+ "training_files": "Data/AbeShinzo20240210/train.list",
33
+ "validation_files": "Data/AbeShinzo20240210/val.list",
34
+ "max_wav_value": 32768.0,
35
+ "sampling_rate": 44100,
36
+ "filter_length": 2048,
37
+ "hop_length": 512,
38
+ "win_length": 2048,
39
+ "n_mel_channels": 128,
40
+ "mel_fmin": 0.0,
41
+ "mel_fmax": null,
42
+ "add_blank": true,
43
+ "n_speakers": 1,
44
+ "cleaned_text": true,
45
+ "spk2id": {
46
+ "AbeShinzo20240210": 0
47
+ },
48
+ "num_styles": 1,
49
+ "style2id": {
50
+ "Neutral": 0
51
+ }
52
+ },
53
+ "model": {
54
+ "use_spk_conditioned_encoder": true,
55
+ "use_noise_scaled_mas": true,
56
+ "use_mel_posterior_encoder": false,
57
+ "use_duration_discriminator": false,
58
+ "use_wavlm_discriminator": true,
59
+ "inter_channels": 192,
60
+ "hidden_channels": 192,
61
+ "filter_channels": 768,
62
+ "n_heads": 2,
63
+ "n_layers": 6,
64
+ "kernel_size": 3,
65
+ "p_dropout": 0.1,
66
+ "resblock": "1",
67
+ "resblock_kernel_sizes": [
68
+ 3,
69
+ 7,
70
+ 11
71
+ ],
72
+ "resblock_dilation_sizes": [
73
+ [
74
+ 1,
75
+ 3,
76
+ 5
77
+ ],
78
+ [
79
+ 1,
80
+ 3,
81
+ 5
82
+ ],
83
+ [
84
+ 1,
85
+ 3,
86
+ 5
87
+ ]
88
+ ],
89
+ "upsample_rates": [
90
+ 8,
91
+ 8,
92
+ 2,
93
+ 2,
94
+ 2
95
+ ],
96
+ "upsample_initial_channel": 512,
97
+ "upsample_kernel_sizes": [
98
+ 16,
99
+ 16,
100
+ 8,
101
+ 2,
102
+ 2
103
+ ],
104
+ "n_layers_q": 3,
105
+ "use_spectral_norm": false,
106
+ "gin_channels": 512,
107
+ "slm": {
108
+ "model": "./slm/wavlm-base-plus",
109
+ "sr": 16000,
110
+ "hidden": 768,
111
+ "nlayers": 13,
112
+ "initial_channel": 64
113
+ }
114
+ },
115
+ "version": "2.2-JP-Extra",
116
+ "model_name": "AbeShinzo20240210"
117
+ }
style_vectors.npy ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:409167fef6fb81edc49212aae7cae1a4ca3ab2ebdd06fabca67a7cd89b741273
3
+ size 5248
style_vectors.npy.bak ADDED
Binary file (1.15 kB). View file