Peter Vickers

This paper proposes a reinforcement learning approach for adapting audio-language models to code-switched speech using group relative policy optimization and verifiable rewards.