Spoken Question Answering on TriviaQA
65.2AccuracyVIRBA
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| VIRBABackbone=Qwen2.5-Omni2025.08 | 65.2 | — | — | |
| VIRBABackbone=Qwen2-Audio2025.08 | 61.8 | — | — | |
| Step-Audio-R12025.08 | 58 | — | — | |
| Kimi-Audio2025.08 | 57.4 | — | — | |
| AF3Full name=Audio Flamingo 32025.08 | 56.2 | — | — | |
| Single-view RL2025.08 | 55.1 | — | — | |
| Qwen2.5-Omni2025.08 | 50.4 | — | — | |
| Fun-Audio-Chat-8BModality=S2S2026.04 | 46.58 | — | — | |
| FreezeEmpathModality=S2S2026.04 | 46.39 | — | — | |
| Kimi-AudioModality=S2S2026.04 | 43.95 | — | — | |
| Lychee-FDType=Native2026.07 | 39.4 | — | — | |
| GLM-4-Voice2025.08 | 39.1 | — | — | |
| Step-Audio2-MiniModality=S2S2026.04 | 38.87 | — | — | |
| LLaMA-Omni2-7BModality=S2S2026.04 | 37.11 | — | — | |
| OpenS2SModality=S2S2026.04 | 31.84 | — | — | |
| Lychee-FD (w/o Sem-Channel)Type=Native2026.07 | 30.1 | — | — | |
| StepAudio-2-miniType=Half-duplex2026.07 | 29.8 | — | — | |
| Fun-Audio-ChatType=Native2026.07 | 27.7 | — | — | |
| VITA 1.5Type=System-level2026.07 | 26 | — | — | |
| SALMONN-omniType=Native2026.07 | 24.2 | — | — | |
| Lychee-FD (w/o Param-Sep)Type=Native2026.07 | 24.2 | — | — | |
| Freeze-OmniType=System-level2026.07 | 23.9 | — | — | |
| Qwen2-Audio base2025.08 | 19.7 | — | — | |
| MoshiType=Native2026.07 | 17.4 | — | — | |
| FLM-audioType=Native2026.07 | 10.4 | — | — | |
| dGSLMType=Native2026.07 | 0.4 | — | — | |
| Baichuan-Omni-1.5Parameters=7B2026.03 | — | 29.51 | 4.35 | |
| Freeze-OmniParameters=7B2026.03 | — | 43.61 | 4.26 | |
| GLM-4-VoiceParameters=9B2026.03 | — | 13.14 | 4.15 | |
| LLaMA-Omni2Parameters=7B2026.03 | — | 6.02 | 4.44 | |
| Qwen2.5-Omni-7BParameters=7B2026.03 | — | 1.47 | 4.29 | |
| SPEECH-OMNI-LITEParameters=8B2026.03 | — | 15.04 | 4.08 | |
| VocalNetParameters=8B2026.03 | — | 4.86 | 4.47 |