Audio Quality Evaluation on Moshi LibriSpeech prompts
2.195VGGish ScoreWMAR
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| WMARh=02026.05 | 2.195 | 0.093 | 2.89 | 3.66 | |
| WMAR (aug)h=02026.05 | 2.153 | 0.113 | 2.98 | 3.69 | |
| WMARh=22026.05 | 2.089 | 0.089 | 3.01 | 3.75 | |
| WMAR (aug)h=12026.05 | 2.036 | 0.109 | 3.19 | 4.06 | |
| WMARh=12026.05 | 2.018 | 0.092 | 3.12 | 4.03 | |
| WMAR (aug)h=22026.05 | 1.985 | 0.096 | 3.1 | 3.78 | |
| Baseh=22026.05 | 1.966 | 0.073 | 3.11 | 3.81 | |
| Baseh=12026.05 | 1.948 | 0.079 | 3.23 | 4.09 | |
| None2026.05 | 1.921 | 0.063 | 3.15 | 3.79 | |
| Oursh=12026.05 | 1.921 | 0.074 | 3.22 | 4.03 | |
| Baseh=02026.05 | 1.858 | 0.072 | 3 | 3.72 | |
| Oursh=22026.05 | 1.823 | 0.059 | 3.15 | 3.8 | |
| Oursh=02026.05 | 1.67 | 0.074 | 3.07 | 3.79 |