MEDIC: Zero-shot Music Editing with Disentangled Inversion Control

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Liu, Huadai, Wang, Jialei, Li, Xiangtai, Wang, Wen, Chen, Qian, Huang, Rongjie, Liu, Yang, Xu, Jiayang, Zhao, Zhou
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914136021008384
author Liu, Huadai
Wang, Jialei
Li, Xiangtai
Wang, Wen
Chen, Qian
Huang, Rongjie
Liu, Yang
Xu, Jiayang
Zhao, Zhou
author_facet Liu, Huadai
Wang, Jialei
Li, Xiangtai
Wang, Wen
Chen, Qian
Huang, Rongjie
Liu, Yang
Xu, Jiayang
Zhao, Zhou
contents Text-guided diffusion models revolutionize audio generation by adapting source audio to specific text prompts. However, existing zero-shot audio editing methods such as DDIM inversion accumulate errors across diffusion steps, reducing the effectiveness. Moreover, existing editing methods struggle with conducting complex non-rigid music edits while maintaining content integrity and high fidelity. To address these challenges, we propose MEDIC, a novel zero-shot music editing system based on innovative Disentangled Inversion Control (DIC) technique, which comprises Harmonized Attention Control and Disentangled Inversion. Disentangled Inversion disentangles the diffusion process into triple branches to rectify the deviated path of the source branch caused by DDIM inversion. Harmonized Attention Control unifies the mutual self-attention control and the cross-attention control with an intermediate Harmonic Branch to progressively generate the desired harmonic and melodic information in the target music. We also introduce ZoME-Bench, a comprehensive music editing benchmark with 1,100 samples covering ten distinct editing categories. ZoME-Bench facilitates both zero-shot and instruction-based music editing tasks. Our method outperforms state-of-the-art inversion techniques in editing fidelity and content preservation. The code and benchmark will be released. Audio samples are available at https://medic-edit.github.io/.
format Preprint
id arxiv_https___arxiv_org_abs_2407_13220
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle MEDIC: Zero-shot Music Editing with Disentangled Inversion Control
Liu, Huadai
Wang, Jialei
Li, Xiangtai
Wang, Wen
Chen, Qian
Huang, Rongjie
Liu, Yang
Xu, Jiayang
Zhao, Zhou
Audio and Speech Processing
Sound
Text-guided diffusion models revolutionize audio generation by adapting source audio to specific text prompts. However, existing zero-shot audio editing methods such as DDIM inversion accumulate errors across diffusion steps, reducing the effectiveness. Moreover, existing editing methods struggle with conducting complex non-rigid music edits while maintaining content integrity and high fidelity. To address these challenges, we propose MEDIC, a novel zero-shot music editing system based on innovative Disentangled Inversion Control (DIC) technique, which comprises Harmonized Attention Control and Disentangled Inversion. Disentangled Inversion disentangles the diffusion process into triple branches to rectify the deviated path of the source branch caused by DDIM inversion. Harmonized Attention Control unifies the mutual self-attention control and the cross-attention control with an intermediate Harmonic Branch to progressively generate the desired harmonic and melodic information in the target music. We also introduce ZoME-Bench, a comprehensive music editing benchmark with 1,100 samples covering ten distinct editing categories. ZoME-Bench facilitates both zero-shot and instruction-based music editing tasks. Our method outperforms state-of-the-art inversion techniques in editing fidelity and content preservation. The code and benchmark will be released. Audio samples are available at https://medic-edit.github.io/.
title MEDIC: Zero-shot Music Editing with Disentangled Inversion Control
topic Audio and Speech Processing
Sound
url https://arxiv.org/abs/2407.13220