---
license: mit
---

# Diff-V2M: A Hierarchical Conditional Diffusion Model with Explicit Rhythmic Modeling for Video-to-Music Generation

<!-- Provide a quick summary of what the model is/does. -->

Here is the training checkpoints of **[Diff-V2M (AAAI'26)](https://arxiv.org/abs/2511.09090)**

## Overview

Diff-V2M is a hierarchical diffusion model with explicit rhythmic modeling and multi-view feature conditioning, achieving state-of-the-art results in video-to-music generation..
<img src="model.jpg" width="770" height="400" alt="model"/>

## Model Sources

<!-- Provide the basic links for the model. -->

- **Repository:** https://github.com/Tayjsl97/Diff-V2M
- **Demo:** [demo page](https://tayjsl97.github.io/Diff-V2M-Demo)

## Citation
If you use our models in your research, please cite it as follows:
```bib
@inproceedings{ji2026diff,
  title={Diff-V2M: A Hierarchical Conditional Diffusion Model with Explicit Rhythmic Modeling for Video-to-Music Generation},
  author={Ji, Shulei and Wang, Zihao and Yu, Jiaxing and Yang, Xiangyuan and Li, Shuyu and Wu, Songruoyao and Zhang, Kejun},
  booktitle={Proceedings of the AAAI Conference on Artificial Intelligence},
  year={2026}
}