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

arXiv: https://arxiv.org/abs/2607.26178 · HTML: https://mlsanity.com/paper/2607.26178/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 20 of its 52 references that are in the map.
- First posted on arXiv 28 Jul 2026.

## Builds on (most cited first, 20 of 20)

- [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025)
- [Just Ask for Calibration: Strategies for Eliciting Calibrated Confidence Scores from Language Models Fine-Tuned with Human Feedback](https://mlsanity.com/paper/2305.14975/) (2023)
- [MultiWOZ -- A Large-Scale Multi-Domain Wizard-of-Oz Dataset for Task-Oriented Dialogue Modelling](https://mlsanity.com/paper/1810.00278/) (2018)
- [Moshi: a speech-text foundation model for real-time dialogue](https://mlsanity.com/paper/2410.00037/) (2024)
- [WhisperX: Time-Accurate Speech Transcription of Long-Form Audio](https://mlsanity.com/paper/2303.00747/) (2023)
- [SODA: Million-scale Dialogue Distillation with Social Commonsense Contextualization](https://mlsanity.com/paper/2212.10465/) (2022)
- [Decoupling Strategy and Generation in Negotiation Dialogues](https://mlsanity.com/paper/1808.09637/) (2018)
- [Full-Duplex-Bench: A Benchmark to Evaluate Full-duplex Spoken Dialogue Models on Turn-taking Capabilities](https://mlsanity.com/paper/2503.04721/) (2025)
- [Beyond Turn-Based Interfaces: Synchronous LLMs as Full-Duplex Dialogue Agents](https://mlsanity.com/paper/2409.15594/) (2024)
- [Language Model Can Listen While Speaking](https://mlsanity.com/paper/2408.02622/) (2024)
- [PersonaPlex: Voice and Role Control for Full Duplex Conversational Speech Models](https://mlsanity.com/paper/2602.06053/) (2026)
- [TurnGPT: a Transformer-based Language Model for Predicting Turn-taking in Spoken Dialog](https://mlsanity.com/paper/2010.10874/) (2020)
- [Talking Turns: Benchmarking Audio Foundation Models on Turn-Taking Dynamics](https://mlsanity.com/paper/2503.01174/) (2025)
- [Scaling Properties of Speech Language Models](https://mlsanity.com/paper/2404.00685/) (2024)
- [SALMONN-omni: A Standalone Speech LLM without Codec Injection for Full-duplex Conversation](https://mlsanity.com/paper/2505.17060/) (2025)
- [Beyond the Turn-Based Game: Enabling Real-Time Conversations with Duplex Models](https://mlsanity.com/paper/2406.15718/) (2024)
- [Large Language Models Know What To Say But Not When To Speak](https://mlsanity.com/paper/2410.16044/) (2024)
- [Predicting Turn-Taking and Backchannel in Human-Machine Conversations Using Linguistic, Acoustic, and Visual Signals](https://mlsanity.com/paper/2505.12654/) (2025)
- [Speechworthy Instruction-tuned Language Models](https://mlsanity.com/paper/2409.14672/) (2024)
- [Beyond Words: Multimodal LLM Knows When to Speak](https://mlsanity.com/paper/2505.14654/) (2025)

## Built on by (most cited first, 1 of 1)

- [Coupled but Late: Turn-Taking Between Full-Duplex Speech Models in Unscripted Dialogue](https://mlsanity.com/paper/2610.08683/) (2026)

## Abstract (arXiv)

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.