Virtual Consistency for Audio Editing
This work addresses a persistent problem in audio editing for users needing efficient, high-quality tools, though it is incremental as it builds on existing neural methods.
The paper tackled the challenge of slow inversion procedures in text-based audio editing by introducing a virtual-consistency based system that adapts diffusion model sampling, achieving substantial speed-ups without quality loss, as shown in benchmarks and a user study with 16 participants.
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.