Audio Editing (Remove) on AudioSet Caps (test)
1.8489LSDHybrid Diffusion Transformer
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Hybrid Diffusion Transformerguidance scale=2.0, sampling steps=2002026.06 | 1.8489 | 0.4289 | 24.3 | 2.94 | 0.96 | 3.2 | |
| RFM-Editingtraining-based=true2026.06 | 1.8613 | 0.4257 | 25.05 | 2.96 | 0.96 | 3.16 | |
| Zero-Shottraining-free=true2026.06 | 1.921 | 0.4798 | 62.9 | 3.9 | 1.12 | 2.79 | |
| AudioEditortraining-free=true2026.06 | 2.3252 | 0.5019 | 36.69 | 4.16 | 1.35 | 3.43 | |
| AUDITtraining-based=true2026.06 | 3.6199 | 0.0878 | 72.44 | 16.46 | 3.51 | 2.83 |