Zero-shot Geometry-Aware Diffusion Guidance for Music Restoration

Jia-Wei Liao, Pin-Chi Pan, Li-Xuan Peng, Sheng-Ping Yang, Yen-Tung Yeh, Cheng-Fu Chou, Yi-Hsuan Yang

Keywords: Music Diffusion Models, Geometry-aware Guidance, Zero-shot Music Restoration

TL;DR: Zero-shot geometry-aware diffusion guidance that enables higher-quality music restoration in a plug-and-play manner.

Abstract: Diffusion models have emerged as powerful generative frameworks and are increasingly used as foundational models for music generation tasks. Recent works have proposed various inference-time optimization methods to adapt pretrained models to downstream tasks. However, these approaches often push noisy samples away from the expected distribution in the diffusion reverse process when applying task-specific loss gradients. To address this issue, we propose Diffusion Geodesic Guidance (DGG), a geometry-aware method that operates on a pretrained diffusion prior preserving the distribution-induced geometry of noisy samples via a closed-form spherical linear interpolation. It updates noisy samples along geodesics of the underlying geometry. We then apply the zero-shot plug-and-play DGG to four multi-task music restoration tasks, achieving consistent improvements over existing training-free baselines and demonstrating a surprisingly wide range of applications for multi-task music restoration.

Track: Paper Track

Confirmation: Paper Track: I confirm that I have followed the formatting guideline and anonymized my submission.

Submission Number: 8