Virtual Consistency for Audio Editing - Audio Samples

Matthieu Cervera, Francesco Paissan, Mirco Ravanelli, Cem Subakan

Welcome to the companion website where you can listen to the edited samples that composed the user study.

Paper Code

Abstract: Free-form, text-based audio editing remains a persistent challenge, despite progress in inversion-based neural methods. Current approaches rely on slow inversion procedures, limiting their practicality. We present a virtual-consistency based audio editing system that bypasses inversion by adapting the sampling process of diffusion models. Our pipeline is model-agnostic, requiring no fine-tuning or architectural changes, and achieves substantial speed-ups over recent neural editing baselines. Crucially, it achieves this efficiency without compromising quality, as demonstrated by quantitative benchmarks and a user study involving 16 participants.

Keywords: Neural Audio Editing, Diffusion Models, Virtual Inversion, Consistency Models