#spoken dialogue

DualTurn: Learning Turn-Taking from Dual-Channel Generative Speech Pretraining

2026-04-03

#Voice AI #Turn-Taking #Full-Duplex #Spoken Dialogue #Speech LLM

Speech-to-speech models know when to speak but can't reason. Cascaded LLM pipelines can reason but only react to silence. DualTurn pretrains on dual-channel human conversation to bring S2S-level turn-taking into a standard ASR-LLM-TTS stack.

DualTurn: Learning Turn-Taking from Dual-Channel Generative Speech Pretraining

2026-03-09 Shangeth Rajaa Interspeech 2026 (Accepted)

#Voice AI #Turn-Taking #Spoken Dialogue #Speech LLM

Dual-channel generative pretraining for learning natural turn-taking in spoken dialogue without labeled data. A 0.5B model that outperforms models 6x its size on turn prediction.

View