Zero-Shot Unsupervised and Text-Based Audio Editing Using DDPM Inversion

TL;DR

This paper introduces two zero-shot audio editing methods using DDPM inversion with pre-trained diffusion models, enabling semantic and text-based modifications of music signals without supervision.

Contribution

It presents the first zero-shot audio editing techniques leveraging DDPM inversion, including a novel unsupervised method for discovering meaningful editing directions.

Findings

Enables semantic audio editing without training data

Demonstrates control over instruments and melody in music signals

Provides open-source samples and code

Abstract

Editing signals using large pre-trained models, in a zero-shot manner, has recently seen rapid advancements in the image domain. However, this wave has yet to reach the audio domain. In this paper, we explore two zero-shot editing techniques for audio signals, which use DDPM inversion with pre-trained diffusion models. The first, which we coin ZEro-shot Text-based Audio (ZETA) editing, is adopted from the image domain. The second, named ZEro-shot UnSupervized (ZEUS) editing, is a novel approach for discovering semantically meaningful editing directions without supervision. When applied to music signals, this method exposes a range of musically interesting modifications, from controlling the participation of specific instruments to improvisations on the melody. Samples and code can be found in https://hilamanor.github.io/AudioEditing/ .