Speech-to-Speech Question-Answering on LlamaQ
76.5AccuracyMinMo
Evaluation Results
| Method | Links | |
|---|---|---|
| MinMoType=AR, S→S capability=true2026.01 | 76.5 | |
| DIFFUSPEECHType=Diff., S→S capability=true2026.01 | 72 | |
| Llama-Omni2Type=AR, S→S capability=true2026.01 | 70.3 | |
| MoshiType=AR, S→S capability=true2026.01 | 68.3 | |
| Lychee-FDType=Native2026.07 | 65.3 | |
| Fun-Audio-ChatType=Native2026.07 | 64.3 | |
| StepAudio-2-miniType=Half-duplex2026.07 | 62 | |
| SALMONN-omniType=Native2026.07 | 61.7 | |
| Lychee-FD (w/o Sem-Channel)Type=Native2026.07 | 61 | |
| MoshiType=Native2026.07 | 54.7 | |
| VITA 1.5Type=System-level2026.07 | 51 | |
| Freeze-OmniType=System-level2026.07 | 50.7 | |
| SpiritLMType=AR, S→S capability=true2026.01 | 40.1 | |
| FLM-audioType=Native2026.07 | 36.7 | |
| Lychee-FD (w/o Param-Sep)Type=Native2026.07 | 36 | |
| SpeechGPTType=AR, S→S capability=true2026.01 | 21.6 | |
| dGSLMType=Native2026.07 | 1.3 |