Hugging Face's logo

lj1995
/
VoiceConversionWebUI

ONNX
Model card Files
xet
 Community
135
lj1995 commited on
Commit
fff0730
·
verified ·
1 Parent(s): e5eacee

Add files using upload-large-folder tool

Browse files
Files changed (9) hide show
  1. pymss_weights/config_mel_band_roformer_karaoke.yaml +71 -0
  2. pymss_weights/dereverb_mel_band_roformer_anvuew.yaml +76 -0
  3. pymss_weights/dereverb_mel_band_roformer_anvuew_sdr_19.1729.ckpt +3 -0
  4. pymss_weights/dereverb_mel_band_roformer_less_aggressive_anvuew_sdr_18.8050.ckpt +3 -0
  5. pymss_weights/model_bs_roformer_ep_317_sdr_12.9755.ckpt +3 -0
  6. pymss_weights/model_bs_roformer_ep_317_sdr_12.9755.yaml +123 -0
  7. pymss_weights/model_bs_roformer_ep_368_sdr_12.9628.ckpt +3 -0
  8. pymss_weights/model_bs_roformer_ep_368_sdr_12.9628.yaml +133 -0
  9. pymss_weights/model_mel_band_roformer_karaoke_aufr33_viperx_sdr_10.1956.ckpt +3 -0
pymss_weights/config_mel_band_roformer_karaoke.yaml ADDED
@@ -0,0 +1,71 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ audio:
2
+ chunk_size: 352800
3
+ dim_f: 1024
4
+ dim_t: 256
5
+ hop_length: 441
6
+ n_fft: 2048
7
+ num_channels: 2
8
+ sample_rate: 44100
9
+ min_mean_abs: 000
10
+
11
+ model:
12
+ dim: 384
13
+ depth: 6
14
+ stereo: true
15
+ num_stems: 1
16
+ time_transformer_depth: 1
17
+ freq_transformer_depth: 1
18
+ num_bands: 60
19
+ dim_head: 64
20
+ heads: 8
21
+ attn_dropout: 0
22
+ ff_dropout: 0
23
+ flash_attn: True
24
+ dim_freqs_in: 1025
25
+ sample_rate: 44100 # needed for mel filter bank from librosa
26
+ stft_n_fft: 2048
27
+ stft_hop_length: 441
28
+ stft_win_length: 2048
29
+ stft_normalized: False
30
+ mask_estimator_depth: 2
31
+ multi_stft_resolution_loss_weight: 1.0
32
+ multi_stft_resolutions_window_sizes: !!python/tuple
33
+ - 4096
34
+ - 2048
35
+ - 1024
36
+ - 512
37
+ - 256
38
+ multi_stft_hop_size: 147
39
+ multi_stft_normalized: False
40
+
41
+ training:
42
+ batch_size: 4
43
+ gradient_accumulation_steps: 1
44
+ grad_clip: 0
45
+ instruments:
46
+ - karaoke
47
+ - other
48
+ lr: 1.0e-05
49
+ patience: 2
50
+ reduce_factor: 0.95
51
+ target_instrument: karaoke
52
+ num_epochs: 1000
53
+ num_steps: 2000
54
+ augmentation: false # enable augmentations by audiomentations and pedalboard
55
+ augmentation_type: null
56
+ use_mp3_compress: false # Deprecated
57
+ augmentation_mix: false # Mix several stems of the same type with some probability
58
+ augmentation_loudness: false # randomly change loudness of each stem
59
+ augmentation_loudness_type: 1 # Type 1 or 2
60
+ augmentation_loudness_min: 0
61
+ augmentation_loudness_max: 0
62
+ q: 0.95
63
+ coarse_loss_clip: false
64
+ ema_momentum: 0.999
65
+ optimizer: adam
66
+ other_fix: false # it's needed for checking on multisong dataset if other is actually instrumental
67
+
68
+ inference:
69
+ batch_size: 1
70
+ dim_t: 256
71
+ num_overlap: 4
pymss_weights/dereverb_mel_band_roformer_anvuew.yaml ADDED
@@ -0,0 +1,76 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ audio:
2
+ chunk_size: 352800
3
+ dim_f: 1024
4
+ dim_t: 256
5
+ hop_length: 441
6
+ n_fft: 2048
7
+ num_channels: 2
8
+ sample_rate: 44100
9
+ min_mean_abs: 0.000
10
+
11
+ model:
12
+ dim: 384
13
+ depth: 6
14
+ stereo: true
15
+ num_stems: 1
16
+ time_transformer_depth: 1
17
+ freq_transformer_depth: 1
18
+ num_bands: 60
19
+ dim_head: 64
20
+ heads: 8
21
+ attn_dropout: 0
22
+ ff_dropout: 0
23
+ flash_attn: True
24
+ dim_freqs_in: 1025
25
+ sample_rate: 44100 # needed for mel filter bank from librosa
26
+ stft_n_fft: 2048
27
+ stft_hop_length: 441
28
+ stft_win_length: 2048
29
+ stft_normalized: False
30
+ mask_estimator_depth: 2
31
+ multi_stft_resolution_loss_weight: 1.0
32
+ multi_stft_resolutions_window_sizes: !!python/tuple
33
+ - 4096
34
+ - 2048
35
+ - 1024
36
+ - 512
37
+ - 256
38
+ multi_stft_hop_size: 147
39
+ multi_stft_normalized: False
40
+
41
+ training:
42
+ batch_size: 3
43
+ gradient_accumulation_steps: 1
44
+ grad_clip: 0
45
+ instruments:
46
+ - noreverb
47
+ - reverb
48
+ lr: 5.0e-05
49
+ patience: 2
50
+ reduce_factor: 0.95
51
+ target_instrument: noreverb
52
+ num_epochs: 1000
53
+ num_steps: 4000
54
+ q: 0.95
55
+ coarse_loss_clip: false
56
+ ema_momentum: 0.999
57
+ optimizer: adamw
58
+ other_fix: true # it's needed for checking on multisong dataset if other is actually instrumental
59
+ use_amp: true # enable or disable usage of mixed precision (float16) - usually it must be true
60
+
61
+ augmentations:
62
+ enable: true # enable or disable all augmentations (to fast disable if needed)
63
+ loudness: true # randomly change loudness of each stem on the range (loudness_min; loudness_max)
64
+ loudness_min: 0.1
65
+ loudness_max: 1.0
66
+ mixup: false # mix several stems of same type with some probability (only works for dataset types: 1, 2, 3)
67
+ mixup_probs: !!python/tuple # 2 additional stems of the same type (1st with prob 0.2, 2nd with prob 0.02)
68
+ - 0.2
69
+ - 0.02
70
+ mixup_loudness_min: 0.5
71
+ mixup_loudness_max: 1.5
72
+
73
+ inference:
74
+ batch_size: 1
75
+ dim_t: 801
76
+ num_overlap: 2
pymss_weights/dereverb_mel_band_roformer_anvuew_sdr_19.1729.ckpt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9262877b87e9ebb0fb808a456b0a411fa677f5df31c8383c1254af531c078970
3
+ size 913107578
pymss_weights/dereverb_mel_band_roformer_less_aggressive_anvuew_sdr_18.8050.ckpt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0db8f1b41c00cead1112e967262a12802fd32e76c0c3a8eb207e772bae25d07b
3
+ size 913107578
pymss_weights/model_bs_roformer_ep_317_sdr_12.9755.ckpt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5b84f37e8d444c8cb30c79d77f613a41c05868ff9c9ac6c7049c00aefae115aa
3
+ size 639331213
pymss_weights/model_bs_roformer_ep_317_sdr_12.9755.yaml ADDED
@@ -0,0 +1,123 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ audio:
2
+ chunk_size: 352800
3
+ dim_f: 1024
4
+ dim_t: 801
5
+ hop_length: 441
6
+ min_mean_abs: 0.0
7
+ n_fft: 2048
8
+ num_channels: 2
9
+ sample_rate: 44100
10
+ inference:
11
+ batch_size: 4
12
+ dim_t: 801
13
+ num_overlap: 2
14
+ model:
15
+ attn_dropout: 0.1
16
+ depth: 12
17
+ dim: 512
18
+ dim_freqs_in: 1025
19
+ dim_head: 64
20
+ ff_dropout: 0.1
21
+ flash_attn: true
22
+ freq_transformer_depth: 1
23
+ freqs_per_bands: !!python/tuple
24
+ - 2
25
+ - 2
26
+ - 2
27
+ - 2
28
+ - 2
29
+ - 2
30
+ - 2
31
+ - 2
32
+ - 2
33
+ - 2
34
+ - 2
35
+ - 2
36
+ - 2
37
+ - 2
38
+ - 2
39
+ - 2
40
+ - 2
41
+ - 2
42
+ - 2
43
+ - 2
44
+ - 2
45
+ - 2
46
+ - 2
47
+ - 2
48
+ - 4
49
+ - 4
50
+ - 4
51
+ - 4
52
+ - 4
53
+ - 4
54
+ - 4
55
+ - 4
56
+ - 4
57
+ - 4
58
+ - 4
59
+ - 4
60
+ - 12
61
+ - 12
62
+ - 12
63
+ - 12
64
+ - 12
65
+ - 12
66
+ - 12
67
+ - 12
68
+ - 24
69
+ - 24
70
+ - 24
71
+ - 24
72
+ - 24
73
+ - 24
74
+ - 24
75
+ - 24
76
+ - 48
77
+ - 48
78
+ - 48
79
+ - 48
80
+ - 48
81
+ - 48
82
+ - 48
83
+ - 48
84
+ - 128
85
+ - 129
86
+ heads: 8
87
+ linear_transformer_depth: 0
88
+ mask_estimator_depth: 2
89
+ multi_stft_hop_size: 147
90
+ multi_stft_normalized: false
91
+ multi_stft_resolution_loss_weight: 1.0
92
+ multi_stft_resolutions_window_sizes: !!python/tuple
93
+ - 4096
94
+ - 2048
95
+ - 1024
96
+ - 512
97
+ - 256
98
+ num_stems: 1
99
+ stereo: true
100
+ stft_hop_length: 441
101
+ stft_n_fft: 2048
102
+ stft_normalized: false
103
+ stft_win_length: 2048
104
+ time_transformer_depth: 1
105
+ training:
106
+ batch_size: 2
107
+ coarse_loss_clip: true
108
+ ema_momentum: 0.999
109
+ grad_clip: 0
110
+ gradient_accumulation_steps: 1
111
+ instruments:
112
+ - vocals
113
+ - other
114
+ lr: 1.0e-05
115
+ num_epochs: 1000
116
+ num_steps: 1000
117
+ optimizer: adam
118
+ other_fix: true
119
+ patience: 2
120
+ q: 0.95
121
+ reduce_factor: 0.95
122
+ target_instrument: vocals
123
+ use_amp: true
pymss_weights/model_bs_roformer_ep_368_sdr_12.9628.ckpt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f6c94864adfb73bbb0ca58ec14d58dd0b364549e9fb61433ae51916f3e2f8d0b
3
+ size 639317465
pymss_weights/model_bs_roformer_ep_368_sdr_12.9628.yaml ADDED
@@ -0,0 +1,133 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ audio:
2
+ chunk_size: 352800
3
+ dim_f: 1024
4
+ dim_t: 801 # don't work (use in model)
5
+ hop_length: 441 # don't work (use in model)
6
+ n_fft: 2048
7
+ num_channels: 2
8
+ sample_rate: 44100
9
+ min_mean_abs: 0.001
10
+
11
+ model:
12
+ dim: 512
13
+ depth: 12
14
+ stereo: true
15
+ num_stems: 1
16
+ time_transformer_depth: 1
17
+ freq_transformer_depth: 1
18
+ freqs_per_bands: !!python/tuple
19
+ - 2
20
+ - 2
21
+ - 2
22
+ - 2
23
+ - 2
24
+ - 2
25
+ - 2
26
+ - 2
27
+ - 2
28
+ - 2
29
+ - 2
30
+ - 2
31
+ - 2
32
+ - 2
33
+ - 2
34
+ - 2
35
+ - 2
36
+ - 2
37
+ - 2
38
+ - 2
39
+ - 2
40
+ - 2
41
+ - 2
42
+ - 2
43
+ - 4
44
+ - 4
45
+ - 4
46
+ - 4
47
+ - 4
48
+ - 4
49
+ - 4
50
+ - 4
51
+ - 4
52
+ - 4
53
+ - 4
54
+ - 4
55
+ - 12
56
+ - 12
57
+ - 12
58
+ - 12
59
+ - 12
60
+ - 12
61
+ - 12
62
+ - 12
63
+ - 24
64
+ - 24
65
+ - 24
66
+ - 24
67
+ - 24
68
+ - 24
69
+ - 24
70
+ - 24
71
+ - 48
72
+ - 48
73
+ - 48
74
+ - 48
75
+ - 48
76
+ - 48
77
+ - 48
78
+ - 48
79
+ - 128
80
+ - 129
81
+ dim_head: 64
82
+ heads: 8
83
+ attn_dropout: 0.1
84
+ ff_dropout: 0.1
85
+ flash_attn: true
86
+ dim_freqs_in: 1025
87
+ stft_n_fft: 2048
88
+ stft_hop_length: 441
89
+ stft_win_length: 2048
90
+ stft_normalized: false
91
+ mask_estimator_depth: 2
92
+ multi_stft_resolution_loss_weight: 1.0
93
+ multi_stft_resolutions_window_sizes: !!python/tuple
94
+ - 4096
95
+ - 2048
96
+ - 1024
97
+ - 512
98
+ - 256
99
+ multi_stft_hop_size: 147
100
+ multi_stft_normalized: False
101
+
102
+ training:
103
+ batch_size: 16
104
+ gradient_accumulation_steps: 1
105
+ grad_clip: 0
106
+ instruments:
107
+ - vocals
108
+ - instrumental
109
+ lr: 5.0e-05
110
+ patience: 2
111
+ reduce_factor: 0.95
112
+ target_instrument: vocals
113
+ num_epochs: 1000
114
+ num_steps: 1000
115
+ augmentation: false # enable augmentations by audiomentations and pedalboard
116
+ augmentation_type: simple1
117
+ use_mp3_compress: false # Deprecated
118
+ augmentation_mix: true # Mix several stems of the same type with some probability
119
+ augmentation_loudness: true # randomly change loudness of each stem
120
+ augmentation_loudness_type: 1 # Type 1 or 2
121
+ augmentation_loudness_min: 0.5
122
+ augmentation_loudness_max: 1.5
123
+ q: 0.95
124
+ coarse_loss_clip: true
125
+ ema_momentum: 0.999
126
+ optimizer: adam
127
+ other_fix: false # it's needed for checking on multisong dataset if other is actually instrumental
128
+ use_amp: true # enable or disable usage of mixed precision (float16) - usually it must be true
129
+
130
+ inference:
131
+ batch_size: 1
132
+ dim_t: 901
133
+ num_overlap: 4
pymss_weights/model_mel_band_roformer_karaoke_aufr33_viperx_sdr_10.1956.ckpt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1de20d459332fe8869aeb01327a31df0032262706e1365114e852dc271779813
3
+ size 913096801