Extending Visual Dynamics for Video-to-Music Generation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Liu, Xiaohao, Tu, Teng, Ma, Yunshan, Chua, Tat-Seng
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916683435737088
author Liu, Xiaohao
Tu, Teng
Ma, Yunshan
Chua, Tat-Seng
author_facet Liu, Xiaohao
Tu, Teng
Ma, Yunshan
Chua, Tat-Seng
contents Music profoundly enhances video production by improving quality, engagement, and emotional resonance, sparking growing interest in video-to-music generation. Despite recent advances, existing approaches remain limited in specific scenarios or undervalue the visual dynamics. To address these limitations, we focus on tackling the complexity of dynamics and resolving temporal misalignment between video and music representations. To this end, we propose DyViM, a novel framework to enhance dynamics modeling for video-to-music generation. Specifically, we extract frame-wise dynamics features via a simplified motion encoder inherited from optical flow methods, followed by a self-attention module for aggregation within frames. These dynamic features are then incorporated to extend existing music tokens for temporal alignment. Additionally, high-level semantics are conveyed through a cross-attention mechanism, and an annealing tuning strategy benefits to fine-tune well-trained music decoders efficiently, therefore facilitating seamless adaptation. Extensive experiments demonstrate DyViM's superiority over state-of-the-art (SOTA) methods.
format Preprint
id arxiv_https___arxiv_org_abs_2504_07594
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Extending Visual Dynamics for Video-to-Music Generation
Liu, Xiaohao
Tu, Teng
Ma, Yunshan
Chua, Tat-Seng
Multimedia
Computer Vision and Pattern Recognition
Music profoundly enhances video production by improving quality, engagement, and emotional resonance, sparking growing interest in video-to-music generation. Despite recent advances, existing approaches remain limited in specific scenarios or undervalue the visual dynamics. To address these limitations, we focus on tackling the complexity of dynamics and resolving temporal misalignment between video and music representations. To this end, we propose DyViM, a novel framework to enhance dynamics modeling for video-to-music generation. Specifically, we extract frame-wise dynamics features via a simplified motion encoder inherited from optical flow methods, followed by a self-attention module for aggregation within frames. These dynamic features are then incorporated to extend existing music tokens for temporal alignment. Additionally, high-level semantics are conveyed through a cross-attention mechanism, and an annealing tuning strategy benefits to fine-tune well-trained music decoders efficiently, therefore facilitating seamless adaptation. Extensive experiments demonstrate DyViM's superiority over state-of-the-art (SOTA) methods.
title Extending Visual Dynamics for Video-to-Music Generation
topic Multimedia
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2504.07594