[Paper] Multi-Faceted Interactivity Alignment in Full-Duplex Speech Models
Source: arXiv - 2606.11167v1
Overview
Full-duplex spoken dialogue models can listen and speak simultaneously, making them a promising architecture for natural conversation. However, current models are trained solely with supervised learning through token-level likelihood maximization, which does not directly optimize interaction-level behaviors, causing interactivity issues such as excessive silence and ill-timed turn-taking. Recent work has applied reinforcement learning (RL) to improve interactivity, but existing methods address only a limited set of interactive behaviors in their rewards. In this work, we propose a post-training alignment method that comprehensively improves the interactivity of full-duplex spoken dialogue models through RL. We address the four canonical axes of interactivity: pause handling, turn-taking, backchanneling, and user interruption. For each axis, we extract short audio segments from human conversation corpora and optimize the model with axis-specific reward functions. An extra LLM-based reward for response quality prevents semantic degradation. We apply our method to two open-source models, Moshi and PersonaPlex, demonstrating consistent improvements in interactivity on both offline evaluation with pre-recorded audio and real-time multi-turn dialogue evaluation.
Key Contributions
This paper presents research in the following areas:
- cs.CL
- eess.AS
Methodology
Please refer to the full paper for detailed methodology.
Practical Implications
This research contributes to the advancement of cs.CL.
Authors
- Atsumoto Ohashi
- Neil Zeghidour
- Alexandre Défossez
- Eugene Kharitonov
Paper Information
- arXiv ID: 2606.11167v1
- Categories: cs.CL, eess.AS
- Published: June 9, 2026
- PDF: Download PDF