src/third_party/MusicSourceSeparationTraining/docs/pretrained_models.md

Pre-trained models

If you trained some good models, please, share them. You can post config and model weights in this issue.

Vocal models

Model Type	Instruments	Metrics (SDR)	Config	Checkpoint
MDX23C	vocals / other	SDR vocals: 10.17	Config	Weights
HTDemucs4 (MVSep finetuned)	vocals / other	SDR vocals: 8.78	Config	Weights
Segm Models (VitLarge23)	vocals / other	SDR vocals: 9.77	Config	Weights
Swin Upernet	vocals / other	SDR vocals: 7.57	Config	Weights
BS Mamba2 (Trained on MUSDB18 only)	vocals / other	SDR vocals: 8.82	Config	Weights
BS Roformer (viperx edition)	vocals / other	SDR vocals: 10.87	Config	Weights
MelBand Roformer (viperx edition)	vocals / other	SDR vocals: 9.67	Config	Weights
MelBand Roformer (KimberleyJensen edition)	vocals / other	SDR vocals: 10.98	Config	Weights

Note: Metrics measured on Multisong Dataset.

Single stem models

Model Type	Instruments	Metrics (SDR)	Config	Checkpoint
HTDemucs4 FT Drums	drums	SDR drums: 11.13	Config	Weights
HTDemucs4 FT Bass	bass	SDR bass: 11.96	Config	Weights
HTDemucs4 FT Other	other	SDR other: 5.85	Config	Weights
HTDemucs4 FT Vocals (Official repository)	vocals	SDR vocals: 8.38	Config	Weights
BS Roformer (viperx edition)	other	SDR other: 6.85	Config	Weights
MelBand Roformer (aufr33 and viperx edition)	crowd	SDR crowd: 5.99	Config	Weights
MelBand Roformer (anvuew edition)	dereverb	---	Config	Weights
MelBand Roformer Denoise (by aufr33)	denoise	---	Config	Weights
MelBand Roformer Denoise Aggressive (by aufr33)	denoise	---	Config	Weights
Apollo LQ MP3 restoration (by JusperLee)	restored	---	Config	Weights
MelBand Roformer Aspiration (by SUC-DriverOld)	aspiration	SDR: 9.85	Config	Weights
MDX23C Phantom Centre extraction (by wesleyr36)	similarity	L1Freq: 72.23	Config	Weights
MelBand Roformer Vocals DeReverb/DeEcho (by SUC-DriverOld)	dry	SDR: 10.01	Config	Weights

Note: All HTDemucs4 FT models output 4 stems, but quality is best only on target stem (all other stems are dummy).

Multi-stem models

Model Type	Instruments	Metrics (SDR)	Config	Checkpoint
BandIt Plus	speech / music / effects	DnR test avg: 11.50 (speech: 15.64, music: 9.18 effects: 9.69)	Config	Weights
HTDemucs4	bass / drums / vocals / other	Multisong avg: 9.16 (bass: 11.76, drums: 10.88 vocals: 8.24 other: 5.74)	Config	Weights
HTDemucs4 (6 stems)	bass / drums / vocals / other / piano / guitar	Multisong (bass: 11.22, drums: 10.22 vocals: 8.05 other: --- piano: --- guitar: ---)	Config	Weights
Demucs3 mmi	bass / drums / vocals / other	Multisong avg: 8.88 (bass: 11.17, drums: 10.70 vocals: 8.22 other: 5.42)	Config	Weights
DrumSep htdemucs (by inagoy)	kick / snare / cymbals / toms	DrumSep test (kick: 10.52, snare: 6.05, toms: 4.68, hh-cymbals: 5.03)	Config	Weights
DrumSep mdx23c (by aufr33 and jarredou)	kick / snare / toms / hh / ride / crash	DrumSep test (kick: 14.54, snare: 9.79, toms: 10.63, hh: 3.19, cymbals (ride + crash): 6.08)	Config	Weights
DrumSep mdx23c (by jarredou)	kick / snare / toms / hh / cymbals	DrumSep test (kick: 16.66, snare: 11.53, toms: 12.33, hh: 4.04, cymbals (ride + crash): 6.36)	Config	Weights

Multi-stem models (MUSDB18HQ)

• Models in this list were trained only on MUSDB18HQ dataset (100 songs train data). These weights are useful for fine-tuning.
• Instruments: bass / drums / vocals / other

Model Type	Metrics (SDR)	Config	Checkpoint
MDX23C	MUSDB test avg: 7.15 (bass: 5.77, drums: 7.93 vocals: 9.23 other: 5.68) Multisong avg: 7.02 (bass: 8.40, drums: 7.73 vocals: 7.36 other: 4.57)	Config	Weights
SCNet Small (by starrytong)	MUSDB test avg: 9.03 (bass: 8.89, drums: 10.44 vocals: 9.90 other: 6.89) Multisong avg: 8.87 (bass: 11.07, drums: 10.79 vocals: 8.27 other: 5.34)	Config	Weights
SCNet Tran Small	MUSDB test avg: 8.92 (bass: 8.07, drums: 10.81 vocals: 9.97 other: 6.84) Multisong avg: 8.97 (bass: 10.99, drums: 10.87 vocals: 8.42 other: 5.63)	Config	Weights
SCNet Masked Small	MUSDB test avg: 8.81 (bass: 8.32, drums: 10.48 vocals: 10.05 other: 6.40) Multisong avg: 8.86 (bass: 11.00, drums: 10.57 vocals: 8.35 other: 5.51)	Config	Weights
SCNet Large	MUSDB test avg: 9.32 (bass: 8.63, drums: 10.89 vocals: 10.69 other: 7.06) Multisong avg: 9.19 (bass: 11.15, drums: 11.04 vocals: 8.94 other: 5.62)	Config	Weights
SCNet Large (by starrytong)	MUSDB test avg: 9.70 (bass: 9.38, drums: 11.15 vocals: 10.94 other: 7.31) Multisong avg: 9.28 (bass: 11.27, drums: 11.23 vocals: 9.05 other: 5.57)	Config	Weights
SCNet XL	MUSDB test avg: 9.80 (bass: 9.23, drums: 11.51 vocals: 11.05 other: 7.41) Multisong avg: 9.72 (bass: 11.87, drums: 11.49 vocals: 9.32 other: 6.19)	Config	Weights
SCNet XL IHF	MUSDB test avg: 10.08 (bass: 9.23, drums: 11.81 vocals: 11.42 other: 7.88) Multisong avg: 9.92 (bass: 11.94, drums: 11.58 vocals: 9.68 other: 6.48)	Config	Weights
SCNet Masked XL IHF	MUSDB test avg: 9.82 (bass: 8.91, drums: 11.62 vocals: 11.06 other: 7.70) Multisong avg: 9.67 (bass: 11.57, drums: 11.38 vocals: 9.43 other: 6.30)	Config	Weights
BS Roformer	MUSDB test avg: 9.65 (bass: 8.48, drums: 11.61 vocals: 11.08 other: 7.44) Multisong avg: 9.38 (bass: 11.08, drums: 11.29 vocals: 9.19 other: 5.96)	Config	Weights
BS Conformer (Medium)	MUSDB test avg: 9.18 (bass: 8.11, drums: 10.96 vocals: 10.63 other: 7.03) Multisong avg: 8.84 (bass: 10.36, drums: 10.73 vocals: 8.75 other: 5.53)	Config	Weights

MelRoformer models

Table of Mel Band Roformers with different paramers