Audio-Visual Speech Separation on VoxCeleb2 (on-screen speakers)
7.71SISDRRAVSS
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| RAVSSON-SCREEN=true, UNSUPERVISED=false2026.02 | 7.71 | 2.32 | 0.64 | 1.92 | 22.5 | 24.9 | |
| SSNAPSON-SCREEN=true, UNSUPERVISED=true2026.02 | 6.17 | 2.16 | 0.65 | 2.78 | 21.7 | 24.4 | |
| SSNAPS w/o L_CT-SSON-SCREEN=true, UNSUPERVISED=true2026.02 | 5.96 | 2.1 | 0.65 | 2.71 | 29.4 | 27.6 | |
| MIXTUREON-SCREEN=true, UNSUPERVISED=N/A2026.02 | -3.08 | 1.44 | 0.33 | 0.86 | 172.3 | 171 |