Add files using upload-large-folder tool

.gitattributes

@@ -157,3 +157,34 @@ MedleyVox-MultiSinger/eval_results/singing_librispeech_iSRNet/examples/ex_72/s0_
 MedleyVox-MultiSinger/eval_results/singing_librispeech_iSRNet/examples/ex_78/s0_estimate.wav filter=lfs diff=lfs merge=lfs -text
 MedleyVox-MultiSinger/eval_results/singing_librispeech_iSRNet/examples/ex_78/s1_estimate.wav filter=lfs diff=lfs merge=lfs -text
 MedleyVox-MultiSinger/eval_results/singing_librispeech_iSRNet/examples/ex_92/mixture.wav filter=lfs diff=lfs merge=lfs -text
+Politrees/UVR_resources/impulse/VS8F-3/301-LargeHall.wav filter=lfs diff=lfs merge=lfs -text
+Politrees/UVR_resources/impulse/VS8F-3/314-Cathedral.wav filter=lfs diff=lfs merge=lfs -text
+Politrees/UVR_resources/impulse/VS8F-3/318-RockSnare.wav filter=lfs diff=lfs merge=lfs -text
+Politrees/UVR_resources/models/Demucs/Demucs_v4/04573f0d-f3cf25b2.th filter=lfs diff=lfs merge=lfs -text
+Politrees/UVR_resources/models/Demucs/Demucs_v3/fa0cb7f9-100d8bf4.th filter=lfs diff=lfs merge=lfs -text
+Politrees/UVR_resources/models/Demucs/Demucs_v4/f7e0c4bc-ba3fe64a.th filter=lfs diff=lfs merge=lfs -text
+Politrees/UVR_resources/impulse/VS8F-1/101-LargeHall.wav filter=lfs diff=lfs merge=lfs -text
+Politrees/UVR_resources/impulse/VS8F-1/103-Strings.wav filter=lfs diff=lfs merge=lfs -text
+Politrees/UVR_resources/impulse/VS8F-1/100-Reverb.wav filter=lfs diff=lfs merge=lfs -text
+Politrees/UVR_resources/models/Demucs/Demucs_v4/d12395a8-e57c48e6.th filter=lfs diff=lfs merge=lfs -text
+Politrees/UVR_resources/impulse/VS8F-1/102-SmallHall.wav filter=lfs diff=lfs merge=lfs -text
+Politrees/UVR_resources/impulse/VS8F-1/110-ShortPlt.wav filter=lfs diff=lfs merge=lfs -text
+Politrees/UVR_resources/impulse/VS8F-1/113-RoomAmb.wav filter=lfs diff=lfs merge=lfs -text
+Politrees/UVR_resources/impulse/VS8F-1/109-CoolPlate.wav filter=lfs diff=lfs merge=lfs -text
+Politrees/UVR_resources/impulse/VS8F-1/106-VocalRoom.wav filter=lfs diff=lfs merge=lfs -text
+Politrees/UVR_resources/impulse/VS8F-1/104-PianoHall.wav filter=lfs diff=lfs merge=lfs -text
+Politrees/UVR_resources/models/Demucs/Demucs_v4/75fc33f5-1941ce65.th filter=lfs diff=lfs merge=lfs -text
+Politrees/UVR_resources/impulse/VS8F-1/105-OrchRoom.wav filter=lfs diff=lfs merge=lfs -text
+Politrees/UVR_resources/impulse/VS8F-1/107-MediumRm.wav filter=lfs diff=lfs merge=lfs -text
+Politrees/UVR_resources/impulse/VS8F-1/112-SoftAmb.wav filter=lfs diff=lfs merge=lfs -text
+Politrees/UVR_resources/impulse/VS8F-1/108-LargeRoom.wav filter=lfs diff=lfs merge=lfs -text
+Politrees/UVR_resources/impulse/VS8F-1/111-VocalPlt.wav filter=lfs diff=lfs merge=lfs -text
+Politrees/UVR_resources/impulse/VS8F-1/118-RockSnare.wav filter=lfs diff=lfs merge=lfs -text
+Politrees/UVR_resources/impulse/VS8F-1/117-RockKick.wav filter=lfs diff=lfs merge=lfs -text
+Politrees/UVR_resources/impulse/VS8F-1/116-GarageDr.wav filter=lfs diff=lfs merge=lfs -text
+Politrees/UVR_resources/impulse/VS8F-1/114-Cathedral.wav filter=lfs diff=lfs merge=lfs -text
+Politrees/UVR_resources/impulse/VS8F-1/115-LongCave.wav filter=lfs diff=lfs merge=lfs -text
+Politrees/UVR_resources/impulse/VS8F-2/200-R1_Reverb1.wav filter=lfs diff=lfs merge=lfs -text
+Politrees/UVR_resources/impulse/VS8F-2/202-R1_SmallHall.wav filter=lfs diff=lfs merge=lfs -text
+Politrees/UVR_resources/impulse/VS8F-2/204-R1_PianoHall.wav filter=lfs diff=lfs merge=lfs -text
+Politrees/UVR_resources/impulse/VS8F-2/207-R1_MediumRm.wav filter=lfs diff=lfs merge=lfs -text

Politrees/UVR_resources/impulse/VS8F-1/100-Reverb.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e0095d99301acc815fafd915bffd6478dfb34ead49cf06e3b53f7d98d382305c
+size 1192672

Politrees/UVR_resources/impulse/VS8F-1/101-LargeHall.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a093f82ab098a1ef004865af68767da88b91a66c418d4f449c268c0d3527ddac
+size 1324768

Politrees/UVR_resources/impulse/VS8F-1/102-SmallHall.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dde33ecb751b0858e722ba9543e37bae0561c45167432173a544690253d8d393
+size 796384

Politrees/UVR_resources/impulse/VS8F-1/103-Strings.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:941d1d3a15a4167bda81598ebdd5c904626a3c1f165fba85c1194bee61740398
+size 1192672

Politrees/UVR_resources/impulse/VS8F-1/104-PianoHall.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b5c675d5549e519b19ff5925253f6efd0cb634c218b399b838ebb30375b59317
+size 1060576

Politrees/UVR_resources/impulse/VS8F-1/105-OrchRoom.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:73ed632ca818fbe3309420ecef4623fdf7df346db53a179b9fe3556031db1a00
+size 664288

Politrees/UVR_resources/impulse/VS8F-1/106-VocalRoom.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5c74a371b9043eef9fa7c806ecc27ec3b949f7990946d3e80fbb4d9d0136c38b
+size 796384

Politrees/UVR_resources/impulse/VS8F-1/107-MediumRm.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1d77584d14207333194a860a9b188117282833928c289ec815184966a6b7163c
+size 424672

Politrees/UVR_resources/impulse/VS8F-1/108-LargeRoom.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9af5c5297ac56d2bd73a970c0028b0c087b854a86fb27a79b054836833cedb57
+size 664288

Politrees/UVR_resources/impulse/VS8F-1/109-CoolPlate.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e220c9e1349889f909d13727e19e898b57a994a585137b9c16d5d6bec6473536
+size 1192672

Politrees/UVR_resources/impulse/VS8F-1/110-ShortPlt.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b80b405c3113144caa136ed2de380d385b9514c7f3df8deae538a8034fbc4e82
+size 842464

Politrees/UVR_resources/impulse/VS8F-1/111-VocalPlt.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8abd4d940b1d3340f0d5e24dd76726e2b138c187a02490e0e57e08c22c1c8fd6
+size 1060576

Politrees/UVR_resources/impulse/VS8F-1/112-SoftAmb.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9a5a1e591dc51a9a5b32b632df8b6c36a7a74372a1ada4daf9eaf91d6a509925
+size 268000

Politrees/UVR_resources/impulse/VS8F-1/113-RoomAmb.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:692dc26f2c18f6ba60834e4f8c9e3ccd58585cc19ec958ec9caad110d7a5ffa0
+size 268000

Politrees/UVR_resources/impulse/VS8F-1/114-Cathedral.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ceb429f76b75df3f39349cea8a1502c766564418dbe575cf393102c81be6b124
+size 1853152

Politrees/UVR_resources/impulse/VS8F-1/115-LongCave.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a98545c4647eecad4d2b15047fa04c4123844ed1eeabc5a0a813f82248ff3dcc
+size 2120416

Politrees/UVR_resources/impulse/VS8F-1/116-GarageDr.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9fde75ec1c9a927001f796611294c3c0e8429b5a3d9c1816ae6bea30efada6ab
+size 268000

Politrees/UVR_resources/impulse/VS8F-1/117-RockKick.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:57dd8a02f8e109c11db81259b8f89abf7e810262e18e39d9020a4fad73dddf4c
+size 326368

Politrees/UVR_resources/impulse/VS8F-1/118-RockSnare.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e1fbcb7539689bae885f9a3a9f23dfe1cdb88cd8be30cc57c15520ffbd1f458b
+size 400096

Politrees/UVR_resources/impulse/VS8F-2/200-R1_Reverb1.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a81291b6857eed249af8e8448a47d5bb15b2afd2e0f79cfaaf1f1025f466f4a9
+size 1192672

Politrees/UVR_resources/impulse/VS8F-2/202-R1_SmallHall.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:917d198e75991fb43cdcb0bca26cc9053df6220859b71264d1cc7cf02ae5551b
+size 796384

Politrees/UVR_resources/impulse/VS8F-2/204-R1_PianoHall.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5492cf3884335da6dcff319511be1814a5fdda07295d7dcac6cbd13696007a8b
+size 1060576

Politrees/UVR_resources/impulse/VS8F-2/207-R1_MediumRm.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:23207fc3a9273ffe1f632215f9a7581e5b39e9ca192c6a3bd8d495bf1f51dd48
+size 464608

Politrees/UVR_resources/impulse/VS8F-3/301-LargeHall.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:72c4b6473a119c3d7e573253591b14de31bff9c25cb8ee8cc6372629930ccb34
+size 1324768

Politrees/UVR_resources/impulse/VS8F-3/314-Cathedral.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:45939810b11ea3f73fdcd760bdacf97f563097354952f4e5c2a56d6f3a2c9b7e
+size 1853152

Politrees/UVR_resources/impulse/VS8F-3/318-RockSnare.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:86032367538db5e717652b58f57ec41ae74e003c96ce6b6434fb94871ffb5d65
+size 596704

Politrees/UVR_resources/models/Bandit/Bandit_Plus/config_dnr_bandit_bsrnn_multi_mus64.yaml

@@ -0,0 +1,78 @@
+name: "MultiMaskMultiSourceBandSplitRNN"
+audio:
+  chunk_size: 264600
+  num_channels: 2
+  sample_rate: 44100
+  min_mean_abs: 0.001
+
+model:
+  in_channel: 1
+  stems: ['speech', 'music', 'effects']
+  band_specs: "musical"
+  n_bands: 64
+  fs: 44100
+  require_no_overlap: false
+  require_no_gap: true
+  normalize_channel_independently: false
+  treat_channel_as_feature: true
+  n_sqm_modules: 8
+  emb_dim: 128
+  rnn_dim: 256
+  bidirectional: true
+  rnn_type: "GRU"
+  mlp_dim: 512
+  hidden_activation: "Tanh"
+  hidden_activation_kwargs: null
+  complex_mask: true
+  n_fft: 2048
+  win_length: 2048
+  hop_length: 512
+  window_fn: "hann_window"
+  wkwargs: null
+  power: null
+  center: true
+  normalized: true
+  pad_mode: "constant"
+  onesided: true
+
+training:
+  batch_size: 4
+  gradient_accumulation_steps: 4
+  grad_clip: 0
+  instruments:
+  - Speech
+  - Music
+  - Effects
+  lr: 9.0e-05
+  patience: 2
+  reduce_factor: 0.95
+  target_instrument: null
+  num_epochs: 1000
+  num_steps: 1000
+  q: 0.95
+  coarse_loss_clip: true
+  ema_momentum: 0.999
+  optimizer: adam
+  other_fix: true # it's needed for checking on multisong dataset if other is actually instrumental
+  use_amp: true # enable or disable usage of mixed precision (float16) - usually it must be true
+
+augmentations:
+  enable: true # enable or disable all augmentations (to fast disable if needed)
+  loudness: true # randomly change loudness of each stem on the range (loudness_min; loudness_max)
+  loudness_min: 0.5
+  loudness_max: 1.5
+  mixup: true # mix several stems of same type with some probability (only works for dataset types: 1, 2, 3)
+  mixup_probs: !!python/tuple # 2 additional stems of the same type (1st with prob 0.2, 2nd with prob 0.02)
+    - 0.2
+    - 0.02
+  mixup_loudness_min: 0.5
+  mixup_loudness_max: 1.5
+  all:
+    channel_shuffle: 0.5 # Set 0 or lower to disable
+    random_inverse: 0.1 # inverse track (better lower probability)
+    random_polarity: 0.5 # polarity change (multiply waveform to -1)
+
+inference:
+  batch_size: 1
+  dim_t: 256
+  num_overlap: 4

Politrees/UVR_resources/models/Bandit/Bandit_v2/config_dnr_bandit_v2_mus64.yaml

@@ -0,0 +1,78 @@
+cls: Bandit
+
+audio:
+  chunk_size: 384000
+  num_channels: 2
+  sample_rate: 48000
+  min_mean_abs: 0.000
+
+kwargs:
+  in_channels: 1
+  stems: ['speech', 'music', 'sfx']
+  band_type: musical
+  n_bands: 64
+  normalize_channel_independently: false
+  treat_channel_as_feature: true
+  n_sqm_modules: 8
+  emb_dim: 128
+  rnn_dim: 256
+  bidirectional: true
+  rnn_type: "GRU"
+  mlp_dim: 512
+  hidden_activation: "Tanh"
+  hidden_activation_kwargs:
+  complex_mask: true
+  use_freq_weights: true
+  n_fft: 2048
+  win_length: 2048
+  hop_length: 512
+  window_fn: "hann_window"
+  wkwargs:
+  power:
+  center: true
+  normalized: true
+  pad_mode: "reflect"
+  onesided: true
+
+training:
+  batch_size: 4
+  gradient_accumulation_steps: 4
+  grad_clip: 0
+  instruments:
+  - Speech
+  - Music
+  - Sfx
+  lr: 9.0e-05
+  patience: 2
+  reduce_factor: 0.95
+  target_instrument:
+  num_epochs: 1000
+  num_steps: 1000
+  q: 0.95
+  coarse_loss_clip: true
+  ema_momentum: 0.999
+  optimizer: adam
+  other_fix: true # it's needed for checking on multisong dataset if other is actually instrumental
+  use_amp: true # enable or disable usage of mixed precision (float16) - usually it must be true
+
+augmentations:
+  enable: true # enable or disable all augmentations (to fast disable if needed)
+  loudness: true # randomly change loudness of each stem on the range (loudness_min; loudness_max)
+  loudness_min: 0.5
+  loudness_max: 1.5
+  mixup: true # mix several stems of same type with some probability (only works for dataset types: 1, 2, 3)
+  mixup_probs: !!python/tuple
+                              # 2 additional stems of the same type (1st with prob 0.2, 2nd with prob 0.02)
+  - 0.2
+  - 0.02
+  mixup_loudness_min: 0.5
+  mixup_loudness_max: 1.5
+  all:
+    channel_shuffle: 0.5 # Set 0 or lower to disable
+    random_inverse: 0.1 # inverse track (better lower probability)
+    random_polarity: 0.5 # polarity change (multiply waveform to -1)
+
+inference:
+  batch_size: 8
+  dim_t: 256
+  num_overlap: 4

Politrees/UVR_resources/models/Demucs/Demucs_v3/UVR_Demucs_Model_1.yaml

@@ -0,0 +1,2 @@
+models: ['ebf34a2db']
+segment: 44

Politrees/UVR_resources/models/Demucs/Demucs_v3/fa0cb7f9-100d8bf4.th

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:100d8bf4dc5e228edfd9bb7a2ec44ad9f44d8f0a845964d38ee8c11ea8ca1399
+size 167405675

Politrees/UVR_resources/models/Demucs/Demucs_v3/mdx.yaml

@@ -0,0 +1,8 @@
+models: ['0d19c1c6', '7ecf8ec1', 'c511e2ab', '7d865c68']
+weights: [
+  [1., 1., 0., 0.],
+  [0., 1., 0., 0.],
+  [1., 0., 1., 1.],
+  [1., 0., 1., 1.],
+]
+segment: 44

Politrees/UVR_resources/models/Demucs/Demucs_v3/mdx_extra.yaml

@@ -0,0 +1,2 @@
+models: ['e51eebcc', 'a1d90b5c', '5d2d6c55', 'cfa93e08']
+segment: 44

Politrees/UVR_resources/models/Demucs/Demucs_v3/mdx_extra_q.yaml

@@ -0,0 +1,2 @@
+models: ['83fc094f', '464b36d7', '14fc6a69', '7fd6ef75']
+segment: 44

Politrees/UVR_resources/models/Demucs/Demucs_v3/mdx_q.yaml

@@ -0,0 +1,8 @@
+models: ['6b9c2ca1', 'b72baf4e', '42e558d4', '305bc58f']
+weights: [
+  [1., 1., 0., 0.],
+  [0., 1., 0., 0.],
+  [1., 0., 1., 1.],
+  [1., 0., 1., 1.],
+]
+segment: 44

Politrees/UVR_resources/models/Demucs/Demucs_v3/repro_mdx_a.yaml

@@ -0,0 +1,2 @@
+models: ['9a6b4851', '1ef250f1', 'fa0cb7f9', '902315c2']
+segment: 44

Politrees/UVR_resources/models/Demucs/Demucs_v3/repro_mdx_a_hybrid_only.yaml

@@ -0,0 +1,2 @@
+models: ['fa0cb7f9', '902315c2', 'fa0cb7f9', '902315c2']
+segment: 44

Politrees/UVR_resources/models/Demucs/Demucs_v3/repro_mdx_a_time_only.yaml

@@ -0,0 +1,2 @@
+models: ['9a6b4851', '9a6b4851', '1ef250f1', '1ef250f1']
+segment: 44

Politrees/UVR_resources/models/Demucs/Demucs_v4/04573f0d-f3cf25b2.th

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f3cf25b222c4eed7cd49dd8b2c9597d50c18bd154090f7b919cfa5f93cf22c49
+size 84141271

Politrees/UVR_resources/models/Demucs/Demucs_v4/75fc33f5-1941ce65.th

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1941ce654b11df4132b9f4eae408556b4c83fad6fe26b4bc0dbcb36b975befb3
+size 167407275

Politrees/UVR_resources/models/Demucs/Demucs_v4/d12395a8-e57c48e6.th

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e57c48e6b0e38af4f7118d7bd08c49f0a0c0edf7d09143bdd902ea0d237303e6
+size 84141271

Politrees/UVR_resources/models/Demucs/Demucs_v4/f7e0c4bc-ba3fe64a.th

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ba3fe64ae8ef66ac9a4857222ce48efbdc5eb3ad375cb79dd13debee5aaa4066
+size 84141271

Politrees/UVR_resources/models/Demucs/Demucs_v4/hdemucs_mmi.yaml

@@ -0,0 +1,2 @@
+models: ['75fc33f5']
+segment: 44

Politrees/UVR_resources/models/Demucs/Demucs_v4/htdemucs.yaml

@@ -0,0 +1 @@
+models: ['955717e8']

Politrees/UVR_resources/models/Demucs/Demucs_v4/htdemucs_6s.yaml

@@ -0,0 +1 @@
+models: ['5c90dfd2']

Politrees/UVR_resources/models/Demucs/Demucs_v4/htdemucs_ft.yaml

@@ -0,0 +1,7 @@
+models: ['f7e0c4bc', 'd12395a8', '92cfc3b6', '04573f0d']
+weights: [
+  [1., 0., 0., 0.],
+  [0., 1., 0., 0.],
+  [0., 0., 1., 0.],
+  [0., 0., 0., 1.],
+]

Politrees/UVR_resources/models/Roformer/MelBand/config_MelBand_Roformer_4stems_XL_v1_by_Aname.yaml

@@ -0,0 +1,167 @@
+audio:
+  chunk_size: 661500
+  dim_f: 1024
+  dim_t: 1101
+  hop_length: 882
+  n_fft: 4096
+  num_channels: 2
+  sample_rate: 44100
+  min_mean_abs: 0.0001
+
+model:
+  dim: 512
+  depth: 12
+  stereo: true
+  num_stems: 4
+  linear_transformer_depth: 0
+  time_transformer_depth: 1
+  freq_transformer_depth: 1
+  num_bands: 60
+  dim_head: 64
+  heads: 8
+  attn_dropout: 0.0
+  ff_dropout: 0.0
+  flash_attn: true
+  dim_freqs_in: 2049
+  sample_rate: 44100  # needed for mel filter bank from librosa
+  stft_n_fft: 4096
+  stft_hop_length: 882
+  stft_win_length: 4096
+  stft_normalized: False
+  mask_estimator_depth: 2
+  multi_stft_resolution_loss_weight: 1.0
+  multi_stft_resolutions_window_sizes: !!python/tuple
+  - 4096
+  - 2048
+  - 1024
+  - 512
+  - 256
+  multi_stft_hop_size: 147
+  multi_stft_normalized: False
+  mlp_expansion_factor: 4
+  use_torch_checkpoint: False # it allows to greatly reduce GPU memory consumption during training (not fully tested)
+  skip_connection: True # Enable skip connection between transformer blocks - can solve problem with gradients and probably faster training
+
+training:
+  batch_size: 1
+  gradient_accumulation_steps: 4
+  grad_clip: 0
+  instruments: ['drums', 'bass', 'other', 'vocals']
+  lr: 2.0e-05
+  patience: 2
+  reduce_factor: 0.95
+  target_instrument: null
+  num_epochs: 1000
+  num_steps: 300
+  q: 0.95
+  coarse_loss_clip: false
+  ema_momentum: 0.999
+  optimizer: adamw
+  read_metadata_procs: 8 # Number of processes to use during metadata reading for dataset. Can speed up metadata generation
+  normalize: false
+  other_fix: false # it's needed for checking on multisong dataset if other is actually instrumental
+  use_amp: true # enable or disable usage of mixed precision (float16) - usually it must be true
+
+augmentations:
+  enable: false # enable or disable all augmentations (to fast disable if needed)
+  loudness: false # randomly change loudness of each stem on the range (loudness_min; loudness_max)
+  loudness_min: 0.5
+  loudness_max: 1.5
+  mixup: true # mix several stems of same type with some probability (only works for dataset types: 1, 2, 3)
+  mixup_probs: !!python/tuple # 2 additional stems of the same type (1st with prob 0.2, 2nd with prob 0.02)
+    - 0.2
+    - 0.02
+    - 0.002
+  mixup_loudness_min: 0.5
+  mixup_loudness_max: 1.5
+
+  # apply mp3 compression to mixture only (emulate downloading mp3 from internet)
+  mp3_compression_on_mixture: 0.01
+  mp3_compression_on_mixture_bitrate_min: 32
+  mp3_compression_on_mixture_bitrate_max: 320
+  mp3_compression_on_mixture_backend: "lameenc"
+
+  all:
+    channel_shuffle: 0.5 # Set 0 or lower to disable
+    random_inverse: 0.01 # inverse track (better lower probability)
+    random_polarity: 0.5 # polarity change (multiply waveform to -1)
+
+  vocals:
+      pitch_shift: 1.0
+      pitch_shift_min_semitones: -12
+      pitch_shift_max_semitones: 12
+      seven_band_parametric_eq: 0.5
+      seven_band_parametric_eq_min_gain_db: -80
+      seven_band_parametric_eq_max_gain_db: 9
+      tanh_distortion: 0.5
+      tanh_distortion_min: 0.1
+      tanh_distortion_max: 1
+      time_stretch: 1.0
+      time_stretch_min_rate: 0.5
+      time_stretch_max_rate: 2
+  bass:
+    pitch_shift: 1.0
+    pitch_shift_min_semitones: -6
+    pitch_shift_max_semitones: 6
+    seven_band_parametric_eq: 0.4
+    seven_band_parametric_eq_min_gain_db: -32
+    seven_band_parametric_eq_max_gain_db: 6
+    tanh_distortion: 1.0
+    tanh_distortion_min: 0.1
+    tanh_distortion_max: 0.5
+    time_stretch: 1.0
+    time_stretch_min_rate: 0.5
+    time_stretch_max_rate: 1.5
+  drums:
+    pitch_shift: 0.1
+    pitch_shift_min_semitones: -6
+    pitch_shift_max_semitones: 6
+    seven_band_parametric_eq: 0.5
+    seven_band_parametric_eq_min_gain_db: -24
+    seven_band_parametric_eq_max_gain_db: 12
+    tanh_distortion: 0.3
+    tanh_distortion_min: 0.1
+    tanh_distortion_max: 0.6
+    time_stretch: 1.0
+    time_stretch_min_rate: 0.333
+    time_stretch_max_rate: 1.5
+  other:
+    pitch_shift: 1.0
+    pitch_shift_min_semitones: -12
+    pitch_shift_max_semitones: 12
+    gaussian_noise: 0.4
+    gaussian_noise_min_amplitude: 0.001
+    gaussian_noise_max_amplitude: 0.15
+    time_stretch: 0.01
+    time_stretch_min_rate: 0.25
+    time_stretch_max_rate: 1.5
+
+inference:
+  batch_size: 1
+  dim_t: 256
+  num_overlap: 4
+  normalize: false
+
+loss_multistft:
+  fft_sizes:
+  - 1024
+  - 2048
+  - 4096
+  hop_sizes:
+  - 147
+  - 256
+  - 512
+  win_lengths:
+  - 1024
+  - 2048
+  - 4096
+  window: "hann_window"
+  scale: "mel"
+  n_bins: 128
+  sample_rate: 44100
+  perceptual_weighting: true
+  w_sc: 16.0
+  w_log_mag: 16.0
+  w_lin_mag: 16.0
+  w_phs: 0.0
+  mag_distance: "L1"

Politrees/UVR_resources/models/Roformer/MelBand/config_melband_roformer_aspiration.yaml

@@ -0,0 +1,76 @@
+audio:
+  chunk_size: 352800
+  dim_f: 1024
+  dim_t: 801 # don't work (use in model)
+  hop_length: 441 # don't work (use in model)
+  n_fft: 2048
+  num_channels: 2
+  sample_rate: 44100
+  min_mean_abs: 0.000
+
+model:
+  dim: 256
+  depth: 8
+  stereo: true
+  num_stems: 2
+  time_transformer_depth: 1
+  freq_transformer_depth: 1
+  num_bands: 60
+  dim_head: 64
+  heads: 8
+  attn_dropout: 0.1
+  ff_dropout: 0.1
+  flash_attn: True
+  dim_freqs_in: 1025
+  sample_rate: 44100  # needed for mel filter bank from librosa
+  stft_n_fft: 2048
+  stft_hop_length: 441
+  stft_win_length: 2048
+  stft_normalized: False
+  mask_estimator_depth: 2
+  multi_stft_resolution_loss_weight: 1.0
+  multi_stft_resolutions_window_sizes: !!python/tuple
+  - 4096
+  - 2048
+  - 1024
+  - 512
+  - 256
+  multi_stft_hop_size: 147
+  multi_stft_normalized: False
+
+training:
+  batch_size: 1
+  gradient_accumulation_steps: 8
+  grad_clip: 0
+  instruments:
+  - aspiration
+  - other
+  lr: 4.0e-05
+  patience: 2
+  reduce_factor: 0.95
+  target_instrument: null
+  num_epochs: 1000
+  num_steps: 1000
+  q: 0.95
+  coarse_loss_clip: true
+  ema_momentum: 0.999
+  optimizer: adam
+  other_fix: false # it's needed for checking on multisong dataset if other is actually instrumental
+  use_amp: true # enable or disable usage of mixed precision (float16) - usually it must be true
+
+augmentations:
+  enable: true # enable or disable all augmentations (to fast disable if needed)
+  loudness: true # randomly change loudness of each stem on the range (loudness_min; loudness_max)
+  loudness_min: 0.5
+  loudness_max: 1.5
+  mixup: false # mix several stems of same type with some probability (only works for dataset types: 1, 2, 3)
+  mixup_probs: !!python/tuple # 2 additional stems of the same type (1st with prob 0.2, 2nd with prob 0.02)
+    - 0.2
+    - 0.02
+  mixup_loudness_min: 0.5
+  mixup_loudness_max: 1.5
+
+inference:
+  batch_size: 4
+  dim_t: 801
+  num_overlap: 2

Politrees/UVR_resources/models/Roformer/MelBand/config_melband_roformer_big_beta4.yaml

@@ -0,0 +1,51 @@
+audio:
+  chunk_size: 485100
+  dim_f: 1024
+  dim_t: 1101
+  hop_length: 441
+  n_fft: 2048
+  num_channels: 2
+  sample_rate: 44100
+  min_mean_abs: 0.000
+
+model:
+  dim: 384
+  depth: 12
+  stereo: true
+  num_stems: 1
+  time_transformer_depth: 1
+  freq_transformer_depth: 1
+  num_bands: 60
+  dim_head: 64
+  heads: 8
+  attn_dropout: 0
+  ff_dropout: 0
+  flash_attn: True
+  dim_freqs_in: 1025
+  sample_rate: 44100  # needed for mel filter bank from librosa
+  stft_n_fft: 2048
+  stft_hop_length: 441
+  stft_win_length: 2048
+  stft_normalized: False
+  mask_estimator_depth: 3
+  multi_stft_resolution_loss_weight: 1.0
+  multi_stft_resolutions_window_sizes: !!python/tuple
+  - 4096
+  - 2048
+  - 1024
+  - 512
+  - 256
+  multi_stft_hop_size: 147
+  multi_stft_normalized: False
+
+training:
+  instruments:
+  - vocals
+  - other
+  target_instrument: vocals
+  use_amp: True
+
+inference:
+  batch_size: 1
+  dim_t: 1101
+  num_overlap: 2

Politrees/UVR_resources/models/Roformer/MelBand/config_melband_roformer_big_beta5e.yaml

@@ -0,0 +1,51 @@
+audio:
+  chunk_size: 485100
+  dim_f: 1024
+  dim_t: 801
+  hop_length: 441
+  n_fft: 2048
+  num_channels: 2
+  sample_rate: 44100
+  min_mean_abs: 0.000
+
+model:
+  dim: 384
+  depth: 6
+  stereo: true
+  num_stems: 1
+  time_transformer_depth: 1
+  freq_transformer_depth: 1
+  num_bands: 60
+  dim_head: 64
+  heads: 8
+  attn_dropout: 0
+  ff_dropout: 0
+  flash_attn: True
+  dim_freqs_in: 1025
+  sample_rate: 44100  # needed for mel filter bank from librosa
+  stft_n_fft: 2048
+  stft_hop_length: 441
+  stft_win_length: 2048
+  stft_normalized: False
+  mask_estimator_depth: 3
+  multi_stft_resolution_loss_weight: 1.0
+  multi_stft_resolutions_window_sizes: !!python/tuple
+  - 4096
+  - 2048
+  - 1024
+  - 512
+  - 256
+  multi_stft_hop_size: 147
+  multi_stft_normalized: False
+
+training:
+  instruments:
+  - vocals
+  - other
+  target_instrument: vocals
+  use_amp: True
+
+inference:
+  batch_size: 1
+  dim_t: 801
+  num_overlap: 2