DuplexGen: Adaptive Synthesis of Human-AI Turn-Taking Dialogues

arXiv cs.CL Papers
dialogue-synthesis human-ai-interaction turn-taking adaptive-synthesis arxiv

Summary

DuplexGen introduces a method for adaptively synthesizing human-AI turn-taking dialogues, addressing the challenge of natural interaction timing in conversational AI.

arXiv:2607.26178v1 Announce Type: new Abstract: Turn-taking is a central component of full-duplex interaction. Which turn-taking behaviors are appropriate varies with the scenario, yet current models apply a single norm regardless of context. This limitation originates in their training data: human-human speech corpora capture natural timing phenomena but provide little role grounding or scenario-specific norms, while heuristic or prompted synthesis methods inject turn-taking behaviors without basing them on human preferences. We introduce DuplexGen, a framework for generating dialogues with scenario-adaptive turn-taking by calibrating LLM predictions against a small set of slot-level human preference annotations. In six cooperative and competitive tasks, human turn-taking preferences differ systematically, and DuplexGen aligns substantially more closely with those preferences than uncalibrated prompting or training solely on generic human-human data; a full-duplex model trained on DuplexGen-generated data exhibits distinctive, human-preferred turn-taking behaviors. These results show that human calibration, not corpus scale or prompt design alone, is what allows turn-taking synthesis to be scenario-specific.
Original Article
View Cached Full Text

Cached at: 07/30/26, 09:56 AM 

# DuplexGen: Adaptive Synthesis of Human–AI Turn-Taking Dialogues
Source: [https://arxiv.org/html/2607.26178](https://arxiv.org/html/2607.26178)
Takyoung Kim1Kang\-wook Kim2,411footnotemark:1Sang Hoon Woo2,5 Julia Hirschberg3Gunhee Kim2Dilek Hakkani\-Tür1

1University of Illinois Urbana\-Champaign2Seoul National University3Columbia University 4University of California, Berkeley5Georgia Institute of Technology [tk30@illinois\.edu](https://arxiv.org/html/2607.26178v1/[email protected])[kangwook@berkeley\.edu](https://arxiv.org/html/2607.26178v1/[email protected])

![[Uncaptioned image]](https://arxiv.org/html/2607.26178v1/all-twemojis.pdf)Project Website:[duplexgen\.github\.io](https://duplexgen.github.io/)

spacing=nonfrench

DuplexGen: Adaptive Synthesis of Human–AI Turn\-Taking Dialogues

Takyoung Kim1††thanks:Equal contribution\.Kang\-wook Kim2,411footnotemark:1Sang Hoon Woo2,5Julia Hirschberg3Gunhee Kim2Dilek Hakkani\-Tür11University of Illinois Urbana\-Champaign2Seoul National University3Columbia University4University of California, Berkeley5Georgia Institute of Technology[tk30@illinois\.edu](https://arxiv.org/html/2607.26178v1/[email protected])[kangwook@berkeley\.edu](https://arxiv.org/html/2607.26178v1/[email protected])![[Uncaptioned image]](https://arxiv.org/html/2607.26178v1/all-twemojis.pdf)Project Website:[duplexgen\.github\.io](https://duplexgen.github.io/)

## References

Similar Articles

Synchronization and Turn-Taking in Full-Duplex Speech Dialogue Models

arXiv cs.CL

This paper analyzes synchronization and turn-taking dynamics in full-duplex speech dialogue models by simulating conversations between two instances of the Moshi model, measuring representational alignment via CKA and predicting turn boundaries with LSTM probes.