Speech Enhancement on StillSonicSet Noise-only condition
1.585PESQStoRM
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| StoRMModel Type=Diffusion2026.03 | 1.585 | 0.648 | 8.522 | 3.418 | 4.026 | 3.105 | 0.93 | |
| Dit-FlowModel Type=Flow Matching2026.03 | 1.575 | 0.612 | 6.001 | 3.461 | 3.916 | 3.174 | 0.936 | |
| SGMSEModel Type=Diffusion2026.03 | 1.314 | 0.497 | 6.158 | 3.444 | 3.519 | 2.949 | 0.895 | |
| Noisy (lower bound)Model Type=-2026.03 | 1.203 | 0.696 | 3.809 | 2.84 | 2.125 | 2.008 | 0.938 |