Noisy Speech Separation on LibriSpeech + FUSS 2 Source Mixtures (test)
18.21SI-SDRSupervised Flow
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Supervised FlowMode=Supervised2026.06 | 18.21 | 0.904 | 3.29 | 2.58 | |
| SURFTeacher=Self-Remixing, Mode=Unsupervised2026.06 | 15.23 | 0.84 | 2.93 | 2.57 | |
| SURFTeacher=ReMixIT, Mode=Unsupervised2026.06 | 14.98 | 0.84 | 2.86 | 2.55 | |
| Self-RemixingMode=Unsupervised2026.06 | 14.81 | 0.784 | 2.81 | 2.56 | |
| ReMixITMode=Unsupervised2026.06 | 14.29 | 0.793 | 2.85 | 2.51 | |
| MixITMode=Unsupervised2026.06 | 14.18 | 0.771 | 2.79 | 2.53 |