SpongeBob: Sync-Aware Harmonious Audio-Visual Generative Editing

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Liang, Sen, Wang, Cong, Guan, Fengbin, Yu, Zhentao, Lu, Yiting, Wang, Yuanzhi, Zhou, Yuan, Li, Xin, Chen, Zhibo
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910272330924032
author Liang, Sen
Wang, Cong
Guan, Fengbin
Yu, Zhentao
Lu, Yiting
Wang, Yuanzhi
Zhou, Yuan
Li, Xin
Chen, Zhibo
author_facet Liang, Sen
Wang, Cong
Guan, Fengbin
Yu, Zhentao
Lu, Yiting
Wang, Yuanzhi
Zhou, Yuan
Li, Xin
Chen, Zhibo
contents Visual and acoustic events in the physical world are inherently coupled, yet existing video editing methods typically adopt decoupled pipelines, lacking bidirectional modality interaction. This results in two key limitations: (i) audio-visual desynchronization and (ii) contextual conflicts between generated audio and preserved content. To address these, we propose SpongeBob, the first end-to-end audio-visual joint editing framework featuring bidirectional cross-modal interaction. For synchronization, a Sync-Aware Mechanism aligns visual edits with sound events via bidirectional attention, temporal alignment, and spatial constraints. For contextual consistency, a Context-Aware Module leverages acoustic and visual context attention to prevent semantic clashes. Additionally, we introduce Sync-Preserving Training and Guidance (SPTG) to enhance alignment without degrading quality. Due to the scarcity of paired data, we construct a scalable data pipeline and a large-scale subject-level dataset. We also propose SpongeBob-Bench for systematic evaluation. Experiments show SpongeBob significantly outperforms existing baselines, improving Sync-C by 30% and Ctx-F1 by 12.5%. Our project page is available at: https://hy-spongebob.github.io/.
format Preprint
id arxiv_https___arxiv_org_abs_2605_25193
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle SpongeBob: Sync-Aware Harmonious Audio-Visual Generative Editing
Liang, Sen
Wang, Cong
Guan, Fengbin
Yu, Zhentao
Lu, Yiting
Wang, Yuanzhi
Zhou, Yuan
Li, Xin
Chen, Zhibo
Computer Vision and Pattern Recognition
Visual and acoustic events in the physical world are inherently coupled, yet existing video editing methods typically adopt decoupled pipelines, lacking bidirectional modality interaction. This results in two key limitations: (i) audio-visual desynchronization and (ii) contextual conflicts between generated audio and preserved content. To address these, we propose SpongeBob, the first end-to-end audio-visual joint editing framework featuring bidirectional cross-modal interaction. For synchronization, a Sync-Aware Mechanism aligns visual edits with sound events via bidirectional attention, temporal alignment, and spatial constraints. For contextual consistency, a Context-Aware Module leverages acoustic and visual context attention to prevent semantic clashes. Additionally, we introduce Sync-Preserving Training and Guidance (SPTG) to enhance alignment without degrading quality. Due to the scarcity of paired data, we construct a scalable data pipeline and a large-scale subject-level dataset. We also propose SpongeBob-Bench for systematic evaluation. Experiments show SpongeBob significantly outperforms existing baselines, improving Sync-C by 30% and Ctx-F1 by 12.5%. Our project page is available at: https://hy-spongebob.github.io/.
title SpongeBob: Sync-Aware Harmonious Audio-Visual Generative Editing
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2605.25193