Audio Quality Evaluation on Moshi conversational audio prompts
0.407VGGish ScoreWMAR
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| WMARh=02026.05 | 0.407 | 0.032 | 3.28 | 4.41 | |
| WMARh=12026.05 | 0.357 | 0.024 | 3.37 | 4.43 | |
| WMARh=22026.05 | 0.336 | 0.03 | 3.3 | 4.39 | |
| WMAR (aug)h=02026.05 | 0.267 | 0.067 | 3.43 | 4.45 | |
| WMAR (aug)h=12026.05 | 0.218 | 0.055 | 3.54 | 4.46 | |
| WMAR (aug)h=22026.05 | 0.189 | 0.062 | 3.47 | 4.43 | |
| Oursh=02026.05 | 0.133 | 0.027 | 3.53 | 4.42 | |
| Baseh=02026.05 | 0.128 | 0.02 | 3.46 | 4.46 | |
| Baseh=22026.05 | 0.111 | 0.021 | 3.5 | 4.44 | |
| Oursh=22026.05 | 0.11 | 0.016 | 3.53 | 4.46 | |
| None2026.05 | 0.08 | 0.023 | 3.54 | 4.43 | |
| Baseh=12026.05 | 0.068 | 0.014 | 3.56 | 4.48 | |
| Oursh=12026.05 | 0.051 | 0.015 | 3.58 | 4.5 |