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

arXiv:2607.26178 · cs.CL · Submitted 2026-07-28 · Read on arXiv

cs.CL

Submitted: 2026-07-28

Updated: 2026-08-29

Comments: EMNLP 2026; Project website: https://duplexgen.github.io

Code: https://github.com/2noise/chattts

License: http://creativecommons.org/licenses/by-sa/4.0/

The gist: Turn-taking is a central component of full-duplex interaction.

Terminology

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.

Sources

Related papers