Hugging Face's logo

Sucial
/
So-VITS4.1_Hanser_model

so-vits-svc
SoVITS
sovits
Model card Files
xet
Community
Sucial commited on
Commit
ec596c0
·
verified ·
1 Parent(s): 06e260e

Upload 24 files

Browse files
Files changed (24) hide show
  1. Hanser_4.1_v1/G_93600.pth +3 -0
  2. Hanser_4.1_v1/config.json +99 -0
  3. Hanser_4.1_v1/config.yaml +48 -0
  4. Hanser_4.1_v1/feature_and_index.pkl +3 -0
  5. Hanser_4.1_v1/kmeans_10000.pt +3 -0
  6. Hanser_4.1_v1/model_30000.pt +3 -0
  7. Hanser_4.1_v2/Hanser_v2.json +105 -0
  8. Hanser_4.1_v2/Hanser_v2.pt +3 -0
  9. Hanser_4.1_v2/Hanser_v2.pth +3 -0
  10. Hanser_4.1_v2/Hanser_v2.yaml +51 -0
  11. Hanser_4.1_v2/feature_and_index_v2.pkl +3 -0
  12. Hanser_4.1_v2/kmeans_10000_v2.pt +3 -0
  13. Hanser_4.1_v3/config.json +107 -0
  14. Hanser_4.1_v3/diffusion.yaml +51 -0
  15. Hanser_4.1_v3/diffusion_44000.pt +3 -0
  16. Hanser_4.1_v3/diffusion_augment.yaml +51 -0
  17. Hanser_4.1_v3/feature_and_index.pkl +3 -0
  18. Hanser_4.1_v3/hanser_ep1716_loss33.06.pth +3 -0
  19. Hanser_4.1_v3/hanser_ep3074_loss30.56.pth +3 -0
  20. Hanser_4.1_v3/hanser_ep4153_loss27.89.pth +3 -0
  21. Hanser_4.1_v3/hanser_ep5272_loss25.09.pth +3 -0
  22. Hanser_4.1_v3/kmeans_10000.pt +3 -0
  23. Hanser_4.1_v3/vocoder_augment/config.json +56 -0
  24. Hanser_4.1_v3/vocoder_augment/hanser.ckpt +3 -0
Hanser_4.1_v1/G_93600.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:84c7403055ccb600626f1e2c22a95815d3914dc1c3ed65927d4d13c83e64f0bb
3
+ size 160937517
Hanser_4.1_v1/config.json ADDED
@@ -0,0 +1,99 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "train": {
3
+ "log_interval": 300,
4
+ "eval_interval": 1200,
5
+ "seed": 1234,
6
+ "epochs": 10000,
7
+ "learning_rate": 0.00005,
8
+ "betas": [
9
+ 0.8,
10
+ 0.99
11
+ ],
12
+ "eps": 1e-09,
13
+ "batch_size": 3,
14
+ "fp16_run": false,
15
+ "lr_decay": 0.999875,
16
+ "segment_size": 10240,
17
+ "init_lr_ratio": 1,
18
+ "warmup_epochs": 0,
19
+ "c_mel": 45,
20
+ "c_kl": 1.0,
21
+ "use_sr": true,
22
+ "max_speclen": 512,
23
+ "port": "8001",
24
+ "keep_ckpts": 3,
25
+ "all_in_mem": false,
26
+ "vol_aug": true
27
+ },
28
+ "data": {
29
+ "training_files": "filelists/train.txt",
30
+ "validation_files": "filelists/val.txt",
31
+ "max_wav_value": 32768.0,
32
+ "sampling_rate": 44100,
33
+ "filter_length": 2048,
34
+ "hop_length": 512,
35
+ "win_length": 2048,
36
+ "n_mel_channels": 80,
37
+ "mel_fmin": 0.0,
38
+ "mel_fmax": 22050
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
+ 2
77
+ ],
78
+ "upsample_initial_channel": 512,
79
+ "upsample_kernel_sizes": [
80
+ 16,
81
+ 16,
82
+ 4,
83
+ 4,
84
+ 4
85
+ ],
86
+ "n_layers_q": 3,
87
+ "use_spectral_norm": false,
88
+ "gin_channels": 768,
89
+ "ssl_dim": 768,
90
+ "n_speakers": 1,
91
+ "vocoder_name": "nsf-hifigan",
92
+ "speech_encoder": "vec768l12",
93
+ "speaker_embedding": false,
94
+ "vol_embedding": true
95
+ },
96
+ "spk": {
97
+ "Hanser": 0
98
+ }
99
+ }
Hanser_4.1_v1/config.yaml ADDED
@@ -0,0 +1,48 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ data:
2
+ block_size: 512
3
+ cnhubertsoft_gate: 10
4
+ duration: 2
5
+ encoder: vec768l12
6
+ encoder_hop_size: 320
7
+ encoder_out_channels: 768
8
+ encoder_sample_rate: 16000
9
+ extensions:
10
+ - wav
11
+ sampling_rate: 44100
12
+ training_files: filelists/train.txt
13
+ validation_files: filelists/val.txt
14
+ device: cuda
15
+ env:
16
+ expdir: logs/44k/diffusion
17
+ gpu_id: 0
18
+ infer:
19
+ method: dpm-solver
20
+ speedup: 10
21
+ model:
22
+ n_chans: 512
23
+ n_hidden: 256
24
+ n_layers: 20
25
+ n_spk: 1
26
+ type: Diffusion
27
+ use_pitch_aug: true
28
+ spk:
29
+ Hanser: 0
30
+ train:
31
+ amp_dtype: fp32
32
+ batch_size: 48
33
+ cache_all_data: true
34
+ cache_device: cpu
35
+ cache_fp16: true
36
+ decay_step: 100000
37
+ epochs: 100000
38
+ gamma: 0.5
39
+ interval_force_save: 10000
40
+ interval_log: 10
41
+ interval_val: 2000
42
+ lr: 0.0002
43
+ num_workers: 2
44
+ save_opt: false
45
+ weight_decay: 0
46
+ vocoder:
47
+ ckpt: pretrain/nsf_hifigan/model
48
+ type: nsf-hifigan
Hanser_4.1_v1/feature_and_index.pkl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c50e7ab7b8b4aff98d37f1469530fae40fa81edef7b3c994bce04e53ecd06aba
3
+ size 152968420
Hanser_4.1_v1/kmeans_10000.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b35818e90e00143c22be23d306a135706ddc4019875125efb404e98514cae76a
3
+ size 31254009
Hanser_4.1_v1/model_30000.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0023d01569e7ceec60c376ced28412fcc7999a3667469c1a42c386883d0330f0
3
+ size 220893767
Hanser_4.1_v2/Hanser_v2.json ADDED
@@ -0,0 +1,105 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "train": {
3
+ "log_interval": 200,
4
+ "eval_interval": 1000,
5
+ "seed": 1234,
6
+ "epochs": 10000,
7
+ "learning_rate": 0.0001,
8
+ "betas": [
9
+ 0.8,
10
+ 0.99
11
+ ],
12
+ "eps": 1e-09,
13
+ "batch_size": 10,
14
+ "fp16_run": false,
15
+ "half_type": "fp16",
16
+ "lr_decay": 0.999875,
17
+ "segment_size": 10240,
18
+ "init_lr_ratio": 1,
19
+ "warmup_epochs": 0,
20
+ "c_mel": 45,
21
+ "c_kl": 1.0,
22
+ "use_sr": true,
23
+ "max_speclen": 512,
24
+ "port": "8001",
25
+ "keep_ckpts": 0,
26
+ "all_in_mem": false,
27
+ "vol_aug": true
28
+ },
29
+ "data": {
30
+ "training_files": "filelists/train.txt",
31
+ "validation_files": "filelists/val.txt",
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": 80,
38
+ "mel_fmin": 0.0,
39
+ "mel_fmax": 22050,
40
+ "unit_interpolate_mode": "nearest"
41
+ },
42
+ "model": {
43
+ "inter_channels": 192,
44
+ "hidden_channels": 192,
45
+ "filter_channels": 768,
46
+ "n_heads": 2,
47
+ "n_layers": 6,
48
+ "kernel_size": 3,
49
+ "p_dropout": 0.1,
50
+ "resblock": "1",
51
+ "resblock_kernel_sizes": [
52
+ 3,
53
+ 7,
54
+ 11
55
+ ],
56
+ "resblock_dilation_sizes": [
57
+ [
58
+ 1,
59
+ 3,
60
+ 5
61
+ ],
62
+ [
63
+ 1,
64
+ 3,
65
+ 5
66
+ ],
67
+ [
68
+ 1,
69
+ 3,
70
+ 5
71
+ ]
72
+ ],
73
+ "upsample_rates": [
74
+ 8,
75
+ 8,
76
+ 2,
77
+ 2,
78
+ 2
79
+ ],
80
+ "upsample_initial_channel": 512,
81
+ "upsample_kernel_sizes": [
82
+ 16,
83
+ 16,
84
+ 4,
85
+ 4,
86
+ 4
87
+ ],
88
+ "n_layers_q": 3,
89
+ "n_flow_layer": 4,
90
+ "use_spectral_norm": false,
91
+ "gin_channels": 768,
92
+ "ssl_dim": 768,
93
+ "n_speakers": 1,
94
+ "vocoder_name": "nsf-hifigan",
95
+ "speech_encoder": "vec768l12",
96
+ "speaker_embedding": false,
97
+ "vol_embedding": true,
98
+ "use_depthwise_conv": false,
99
+ "flow_share_parameter": false,
100
+ "use_automatic_f0_prediction": true
101
+ },
102
+ "spk": {
103
+ "Hanser": 0
104
+ }
105
+ }
Hanser_4.1_v2/Hanser_v2.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c9d7617ebdc39d5b3384d76c158ec65a8971241e583ad0871e434c0cf67428b5
3
+ size 220894208
Hanser_4.1_v2/Hanser_v2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ad1cf2f3d57a5ec8808ecc9be7962e88953df88127b46733b960cd9f4afad020
3
+ size 160937581
Hanser_4.1_v2/Hanser_v2.yaml ADDED
@@ -0,0 +1,51 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ data:
2
+ block_size: 512
3
+ cnhubertsoft_gate: 10
4
+ duration: 2
5
+ encoder: vec768l12
6
+ encoder_hop_size: 320
7
+ encoder_out_channels: 768
8
+ encoder_sample_rate: 16000
9
+ extensions:
10
+ - wav
11
+ sampling_rate: 44100
12
+ training_files: filelists/train.txt
13
+ unit_interpolate_mode: nearest
14
+ validation_files: filelists/val.txt
15
+ device: cuda
16
+ env:
17
+ expdir: logs/44k/diffusion
18
+ gpu_id: 0
19
+ infer:
20
+ method: dpm-solver++
21
+ speedup: 10
22
+ model:
23
+ k_step_max: 0
24
+ n_chans: 512
25
+ n_hidden: 256
26
+ n_layers: 20
27
+ n_spk: 1
28
+ timesteps: 1000
29
+ type: Diffusion
30
+ use_pitch_aug: true
31
+ spk:
32
+ Hanser: 0
33
+ train:
34
+ amp_dtype: fp32
35
+ batch_size: 48
36
+ cache_all_data: true
37
+ cache_device: cpu
38
+ cache_fp16: true
39
+ decay_step: 100000
40
+ epochs: 100000
41
+ gamma: 0.5
42
+ interval_force_save: 5000
43
+ interval_log: 10
44
+ interval_val: 2000
45
+ lr: 0.0001
46
+ num_workers: 4
47
+ save_opt: false
48
+ weight_decay: 0
49
+ vocoder:
50
+ ckpt: pretrain/nsf_hifigan/model
51
+ type: nsf-hifigan
Hanser_4.1_v2/feature_and_index_v2.pkl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ae042eccbb67ad82b9ba43836b622167212f6eb6a071e89e9b25d877c38da3ad
3
+ size 499915020
Hanser_4.1_v2/kmeans_10000_v2.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:51df50819b3dd8982781ec039836dee6856ce40fe58b8afa983c43ca0700dd87
3
+ size 32327668
Hanser_4.1_v3/config.json ADDED
@@ -0,0 +1,107 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "train": {
3
+ "log_interval": 200,
4
+ "eval_interval": 800,
5
+ "seed": 1234,
6
+ "epochs": 10000,
7
+ "learning_rate": 0.0001,
8
+ "betas": [
9
+ 0.8,
10
+ 0.99
11
+ ],
12
+ "eps": 1e-09,
13
+ "batch_size": 16,
14
+ "fp16_run": false,
15
+ "half_type": "bf16",
16
+ "lr_decay": 0.999875,
17
+ "segment_size": 10240,
18
+ "init_lr_ratio": 1,
19
+ "warmup_epochs": 0,
20
+ "c_mel": 45,
21
+ "c_kl": 1.0,
22
+ "use_sr": true,
23
+ "max_speclen": 512,
24
+ "port": "8001",
25
+ "keep_ckpts": 3,
26
+ "all_in_mem": false,
27
+ "vol_aug": true
28
+ },
29
+ "data": {
30
+ "training_files": "filelists/train.txt",
31
+ "validation_files": "filelists/val.txt",
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": 80,
38
+ "mel_fmin": 0.0,
39
+ "mel_fmax": 22050,
40
+ "unit_interpolate_mode": "nearest"
41
+ },
42
+ "model": {
43
+ "inter_channels": 192,
44
+ "hidden_channels": 192,
45
+ "filter_channels": 768,
46
+ "n_heads": 2,
47
+ "n_layers": 6,
48
+ "kernel_size": 3,
49
+ "p_dropout": 0.1,
50
+ "resblock": "1",
51
+ "resblock_kernel_sizes": [
52
+ 3,
53
+ 7,
54
+ 11
55
+ ],
56
+ "resblock_dilation_sizes": [
57
+ [
58
+ 1,
59
+ 3,
60
+ 5
61
+ ],
62
+ [
63
+ 1,
64
+ 3,
65
+ 5
66
+ ],
67
+ [
68
+ 1,
69
+ 3,
70
+ 5
71
+ ]
72
+ ],
73
+ "upsample_rates": [
74
+ 8,
75
+ 8,
76
+ 2,
77
+ 2,
78
+ 2
79
+ ],
80
+ "upsample_initial_channel": 512,
81
+ "upsample_kernel_sizes": [
82
+ 16,
83
+ 16,
84
+ 4,
85
+ 4,
86
+ 4
87
+ ],
88
+ "n_layers_q": 3,
89
+ "n_layers_trans_flow": 3,
90
+ "n_flow_layer": 4,
91
+ "use_spectral_norm": false,
92
+ "gin_channels": 768,
93
+ "ssl_dim": 768,
94
+ "n_speakers": 1,
95
+ "vocoder_name": "nsf-hifigan",
96
+ "speech_encoder": "vec768l12",
97
+ "speaker_embedding": false,
98
+ "vol_embedding": true,
99
+ "use_depthwise_conv": false,
100
+ "flow_share_parameter": false,
101
+ "use_automatic_f0_prediction": true,
102
+ "use_transformer_flow": false
103
+ },
104
+ "spk": {
105
+ "hanser": 0
106
+ }
107
+ }
Hanser_4.1_v3/diffusion.yaml ADDED
@@ -0,0 +1,51 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ data:
2
+ block_size: 512
3
+ cnhubertsoft_gate: 10
4
+ duration: 2
5
+ encoder: vec768l12
6
+ encoder_hop_size: 320
7
+ encoder_out_channels: 768
8
+ encoder_sample_rate: 16000
9
+ extensions:
10
+ - wav
11
+ sampling_rate: 44100
12
+ training_files: filelists/train.txt
13
+ unit_interpolate_mode: nearest
14
+ validation_files: filelists/val.txt
15
+ device: cuda
16
+ env:
17
+ expdir: logs/44k/diffusion
18
+ gpu_id: 0
19
+ infer:
20
+ method: dpm-solver++
21
+ speedup: 10
22
+ model:
23
+ k_step_max: 0
24
+ n_chans: 512
25
+ n_hidden: 256
26
+ n_layers: 20
27
+ n_spk: 1
28
+ timesteps: 1000
29
+ type: Diffusion
30
+ use_pitch_aug: true
31
+ spk:
32
+ hanser: 0
33
+ train:
34
+ amp_dtype: fp32
35
+ batch_size: 48
36
+ cache_all_data: true
37
+ cache_device: cpu
38
+ cache_fp16: true
39
+ decay_step: 100000
40
+ epochs: 100000
41
+ gamma: 0.5
42
+ interval_force_save: 5000
43
+ interval_log: 10
44
+ interval_val: 2000
45
+ lr: 0.0001
46
+ num_workers: 4
47
+ save_opt: false
48
+ weight_decay: 0
49
+ vocoder:
50
+ ckpt: pretrain/nsf_hifigan/model
51
+ type: nsf-hifigan
Hanser_4.1_v3/diffusion_44000.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:310322606ac13439d13636c7658eb6b471cacabee0fe1cfb476c82e0c24b5bb1
3
+ size 220893767
Hanser_4.1_v3/diffusion_augment.yaml ADDED
@@ -0,0 +1,51 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ data:
2
+ block_size: 512
3
+ cnhubertsoft_gate: 10
4
+ duration: 2
5
+ encoder: vec768l12
6
+ encoder_hop_size: 320
7
+ encoder_out_channels: 768
8
+ encoder_sample_rate: 16000
9
+ extensions:
10
+ - wav
11
+ sampling_rate: 44100
12
+ training_files: filelists/train.txt
13
+ unit_interpolate_mode: nearest
14
+ validation_files: filelists/val.txt
15
+ device: cuda
16
+ env:
17
+ expdir: logs/44k/diffusion
18
+ gpu_id: 0
19
+ infer:
20
+ method: dpm-solver++
21
+ speedup: 10
22
+ model:
23
+ k_step_max: 0
24
+ n_chans: 512
25
+ n_hidden: 256
26
+ n_layers: 20
27
+ n_spk: 1
28
+ timesteps: 1000
29
+ type: Diffusion
30
+ use_pitch_aug: true
31
+ spk:
32
+ hanser: 0
33
+ train:
34
+ amp_dtype: fp32
35
+ batch_size: 48
36
+ cache_all_data: true
37
+ cache_device: cpu
38
+ cache_fp16: true
39
+ decay_step: 100000
40
+ epochs: 100000
41
+ gamma: 0.5
42
+ interval_force_save: 5000
43
+ interval_log: 10
44
+ interval_val: 2000
45
+ lr: 0.0001
46
+ num_workers: 4
47
+ save_opt: false
48
+ weight_decay: 0
49
+ vocoder:
50
+ ckpt: pretrain/hanser_vocoder/hanser.ckpt
51
+ type: nsf-hifigan
Hanser_4.1_v3/feature_and_index.pkl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9429d9096b9726f79bf4c2d0f2f582fd0ee9da2555e3b8be3354b5794c04b4b0
3
+ size 31588700
Hanser_4.1_v3/hanser_ep1716_loss33.06.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cd5f2c4630c4024114e6ad2dc23d0dcfa2ad8373602f544431693753c48b5d40
3
+ size 160942069
Hanser_4.1_v3/hanser_ep3074_loss30.56.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1c5125e2c0ab1c32712925a9fd5261548171684fd2dadddcf332abf2a6441a21
3
+ size 160942721
Hanser_4.1_v3/hanser_ep4153_loss27.89.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2a250c4963cc10533294d21dab1c96057cc6b45d36ac267b94c9eacb0adfa4d5
3
+ size 160942721
Hanser_4.1_v3/hanser_ep5272_loss25.09.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2380f828c50a9be03fde5d389db1ee5a792eeff0d2fa9bb31c423c8eb3dbcc0f
3
+ size 160942721
Hanser_4.1_v3/kmeans_10000.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b36d49552f154352446b45f7b430f0f3e327a2c1c6a4aa94f72251c0966c67ff
3
+ size 31268281
Hanser_4.1_v3/vocoder_augment/config.json ADDED
@@ -0,0 +1,56 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "discriminator_periods": [
3
+ 3,
4
+ 5,
5
+ 7,
6
+ 11,
7
+ 17,
8
+ 23,
9
+ 37
10
+ ],
11
+ "resblock": "1",
12
+ "resblock_dilation_sizes": [
13
+ [
14
+ 1,
15
+ 3,
16
+ 5
17
+ ],
18
+ [
19
+ 1,
20
+ 3,
21
+ 5
22
+ ],
23
+ [
24
+ 1,
25
+ 3,
26
+ 5
27
+ ]
28
+ ],
29
+ "resblock_kernel_sizes": [
30
+ 3,
31
+ 7,
32
+ 11
33
+ ],
34
+ "upsample_initial_channel": 512,
35
+ "upsample_kernel_sizes": [
36
+ 16,
37
+ 16,
38
+ 4,
39
+ 4,
40
+ 4
41
+ ],
42
+ "upsample_rates": [
43
+ 8,
44
+ 8,
45
+ 2,
46
+ 2,
47
+ 2
48
+ ],
49
+ "sampling_rate": 44100,
50
+ "num_mels": 128,
51
+ "hop_size": 512,
52
+ "n_fft": 2048,
53
+ "win_size": 2048,
54
+ "fmin": 40,
55
+ "fmax": 16000
56
+ }
Hanser_4.1_v3/vocoder_augment/hanser.ckpt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2b22c3b675a8b6628ea1ad82ff187b45b96affc610f6a6d26894a08bbdde0420
3
+ size 56813357