PrismAudio: Decomposed Chain-of-Thoughts and Multi-dimensional Rewards for Video-to-Audio Generation

TL;DR

PrismAudio introduces a novel reinforcement learning framework with specialized chain-of-thought modules and multi-dimensional rewards for improved video-to-audio generation, addressing objective entanglement and enhancing perceptual quality.

Contribution

It is the first to integrate RL with decomposed reasoning modules and tailored rewards for multidimensional video-to-audio generation, improving interpretability and performance.

Findings

Achieves state-of-the-art results across perceptual dimensions

Introduces a new balanced benchmark for V2A tasks

Reduces training overhead with hybrid ODE-SDE sampling

Abstract

Video-to-Audio (V2A) generation requires balancing four critical perceptual dimensions: semantic consistency, audio-visual temporal synchrony, aesthetic quality, and spatial accuracy; yet existing methods suffer from objective entanglement that conflates competing goals in single loss functions and lack human preference alignment. We introduce PrismAudio, the first framework to integrate Reinforcement Learning into V2A generation with specialized Chain-of-Thought (CoT) planning. Our approach decomposes monolithic reasoning into four specialized CoT modules (Semantic, Temporal, Aesthetic, and Spatial CoT), each paired with targeted reward functions. This CoT-reward correspondence enables multidimensional RL optimization that guides the model to jointly generate better reasoning across all perspectives, solving the objective entanglement problem while preserving interpretability. To make this optimization computationally practical, we propose Fast-GRPO, which employs hybrid ODE-SDE sampling that dramatically reduces the training overhead compared to existing GRPO implementations. We also introduce AudioCanvas, a rigorous benchmark that is more distributionally balanced and covers more realistically diverse and challenging scenarios than existing datasets, with 300 single-event classes and 501 multi-event samples. Experimental results demonstrate that PrismAudio achieves state-of-the-art performance across all four perceptual dimensions on both the in-domain VGGSound test set and out-of-domain AudioCanvas benchmark. The project page is available at https://PrismAudio.github.io.