Speech Separation on WHAM!
18.4SI-SNRi (dB)SepReformer-L
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| SepReformer-LDynamic Mixing=true2024.06 | 18.4 | 18.7 | |
| Mossformer2(L)Dynamic Mixing=true2024.06 | 18.1 | — | |
| Mossformer(L)Dynamic Mixing=true2024.06 | 17.3 | — | |
| ISCIT+Dynamic Mixing=true, Additional speaker information=true2024.06 | 16.9 | 17.2 | |
| SepformerDynamic Mixing=true2024.06 | 16.4 | 16.7 | |
| WavesplitDynamic mixing=true2020.02 | 16 | 16.5 | |
| WavesplitDynamic mixing=false2020.02 | 15.4 | 15.8 | |
| MULCAT2020.02 | 15.2 | — | |
| DPRNN2020.02 | 13.9 | — | |
| Learnable fbank2020.02 | 12.9 | — | |
| Learnable filter bank2020.02 | 12.9 | — | |
| Conv-TasNet2020.02 | 12.7 | — | |
| Conv-TasNet2020.02 | 12.7 | — | |
| BLSTM-TasNet2020.02 | 12 | — | |
| Chimera++2020.02 | 9.9 | — |