Spoken Question Answering on LlamaQ S→T
75.7AccuracyVITA 1.5
Evaluation Results
| Method | Links | |
|---|---|---|
| VITA 1.5Type=System-level2026.07 | 75.7 | |
| StepAudio-2-miniType=Half-duplex2026.07 | 74.7 | |
| Lychee-FDType=Native2026.07 | 73.7 | |
| Fun-Audio-ChatType=Native2026.07 | 72.3 | |
| Freeze-OmniType=System-level2026.07 | 71.3 | |
| Lychee-FD (w/o Sem-Channel)Type=Native2026.07 | 69.3 | |
| SALMONN-omniType=Native2026.07 | 67 | |
| Lychee-FD (w/o Param-Sep)Type=Native2026.07 | 67 | |
| MoshiType=Native2026.07 | 62.3 | |
| FLM-audioType=Native2026.07 | 41.3 |