X$^3$-OPD: Distilling Reasoning into Large Audio-Language Models via On-Policy Alignment

Dongjie Fu, Di Cao, Xize Cheng, Zihan Zhang, Wenxu Jia et al.|July 23, 2026arXiv

Key Takeaway

Audio-language models can learn complex reasoning by generating their own reasoning trajectories while receiving token-level guidance from text teachers, without requiring massive amounts of audio reasoning data.

Summary

This paper presents X³-OPD, a method for teaching audio-language models to reason better by learning from text-based models.

training multimodal reasoning

Key Terms

on-policy-learning cross-modal-distillation chain-of-thought audio-language-model token-level-guidance