New paper "MultiTalk" extends the Moshi paradigm to long, multi-party, bilingual conversations, releasing 57.6k hours of speech data and codec-frame-level full-duplex modeling for English and Chinese. Aims to address…

#OpenSourceAI #SpeechModels #Moshi #FullDuplex
https://arxiv.org/abs/2609.36903

With full-duplex voice everywhere this month, the bar moved to turn-taking: ~800ms before an agent feels like a walkie-talkie. What's the most lag you'll tolerate on a call? #VoiceAI #FullDuplex