Music Source Separation On Musdb18 Hq

Métriques

SDR (avg)

SDR (bass)

SDR (drums)

SDR (others)

SDR (vocals)

Résultats

Résultats de performance de divers modèles sur ce benchmark

						Paper Title
BS-RoFormer (L=12, OA)	11.99	13.32	12.91	9.01	12.72	-
BS-RoFormer (L=6, OA)	9.80	11.31	9.49	7.73	10.66	-
SCNet-large	9.69	9.49	10.98	7.44	10.86	SCNet: Sparse Compression Network for Music Source Separation
Sparse HT Demucs (fine tuned)	9.20	10.47	10.83	6.41	9.37	Hybrid Transformers for Music Source Separation
SCNet	9.00	8.82	10.51	6.76	9.89	SCNet: Sparse Compression Network for Music Source Separation
Hybrid Transformer Demucs (f.t.)	9.00	10.39	10.08	6.32	9.20	Hybrid Transformers for Music Source Separation
Band-Split RNN (semi-sup.)	8.97	8.16	10.15	7.08	10.47	Music Source Separation with Band-split RNN
TFC-TDF-UNet (v3)	8.34	8.45	8.44	6.86	9.59	Sound Demixing Challenge 2023 Music Demixing Track Technical Report: TFC-TDF-UNet v3
Band-Split RNN	8.24	7.22	9.01	6.70	10.01	Music Source Separation with Band-split RNN
Dual-Path TFC-TDF UNet (DTTNet)	8.15	7.55	7.82	7.02	10.21	Music Source Separation Based on a Lightweight Deep Learning Framework (DTTNET: DUAL-PATH TFC-TDF UNET)
Hybrid Demucs	7.68	8.76	8.24	5.59	8.13	Hybrid Spectrogram and Waveform Source Separation
KUIELab-MDX-Net	7.47	7.83	7.20	5.90	8.97	KUIELab-MDX-Net: A Two-Stream Neural Network for Music Demixing
CWS-PResUNet	6.77	5.93	6.38	5.84	8.92	CWS-PResUNet: Music Source Separation with Channel-wise Subband Phase-aware ResUNet
Unmix	4.188	4.073	4.925	2.695	5.06	Transfer Learning with Jukebox for Music Source Separation

0 of 14 row(s) selected.