Non-intrusive Speech Quality Assessment on AudioMOS (test)
0.282UTT MSEbaseline
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| baselineTrain Data=AudioMOS_train2026.02 | 0.282 | 0.83 | 0.678 | 0.138 | 0.961 | 0.852 | |
| SA-SSL-MOSTrain Data=AudioMOS_train2026.02 | 0.375 | 0.83 | 0.679 | 0.286 | 0.953 | 0.826 | |
| SA-SSL-MOSTrain Data=NISQA+AudioMOS_train2026.02 | 0.377 | 0.848 | 0.75 | 0.323 | 0.943 | 0.856 | |
| baselineTrain Data=NISQA+AudioMOS_train2026.02 | 0.465 | 0.819 | 0.731 | 0.385 | 0.936 | 0.845 | |
| SA-SSL-MOSTrain Data=NISQA2026.02 | 0.555 | 0.789 | 0.721 | 0.424 | 0.911 | 0.754 | |
| baselineTrain Data=NISQA2026.02 | 0.835 | 0.798 | 0.712 | 0.641 | 0.92 | 0.781 |