List your company

Speech-to-Speech Question-Answering on Average (LlamaQ, TriviaQA, WebQ)

50.6AccuracyDIFFUSPEECH
4.63216.56628.540.434Jan 30, 2026

Evaluation results

MethodLinks
DIFFUSPEECHType=Diff., S→S capability=true2026.0150.6
MoshiType=AR, S→S capability=true2026.0145.1
MinMoType=AR, S→S capability=true2026.0143.1
Llama-Omni2Type=AR, S→S capability=true2026.0140.6
SpiritLMType=AR, S→S capability=true2026.0117.5
SpeechGPTType=AR, S→S capability=true2026.016.4