metadata
license: mit
task_categories:
- text-to-audio
tags:
- joint-audio-video-generation
- multimodal
- sounding-video
JavisDiT++ Datasets
Project Page | Paper | GitHub
This repository contains data associated with JavisDiT++, a concise yet powerful framework for unified modeling and optimization of Joint Audio-Video Generation (JAVG). It produces synchronized and semantically aligned sound and vision from textual descriptions.
Dataset Description
The JavisDiT project releases several data components:
- JavisBench: A comprehensive benchmark for evaluating joint audio-video generation across quality, consistency, and synchrony.
- JavisData-Audio: Audio pre-training data used to initialize text-to-audio generation.
Data Structure
Training and evaluation entries are managed using .csv files containing metadata such as video/audio paths, number of frames, resolution, and textual descriptions.
| Column | Description |
|---|---|
path |
Path to the video file |
id |
Unique identifier |
num_frames |
Total frames |
audio_path |
Path to the corresponding audio file |
text |
Textual description/prompt |
Usage
You can download the benchmark data or pre-processed audio dataset using the Hugging Face CLI:
Download JavisBench
hf download --repo-type dataset JavisVerse/JavisBench --local-dir data/eval/JavisBench
Download JavisData-Audio
hf download --repo-type dataset JavisVerse/JavisData-Audio --local-dir /path/to/audio
Citation
If you find JavisDiT++ useful in your research, please cite the following papers:
@inproceedings{liu2026javisdit++,
title = {JavisDiT++: Unified Modeling and Optimization for Joint Audio-Video Generation},
author = {Liu, Kai and Zheng, Yanhao and Wang, Kai and Wu, Shengqiong and Zhang, Rongjunchen and Luo, Jiebo and Hatzinakos, Dimitrios and Liu, Ziwei and Fei, Hao and Chua, Tat-Seng},
conference = {The Fourteenth International Conference on Learning Representations},
year = {2026},
}
@inproceedings{liu2025javisdit,
title = {JavisDiT: Joint Audio-Video Diffusion Transformer with Hierarchical Spatio-Temporal Prior Synchronization},
author = {Liu, Kai and Li, Wei and Chen, Lai and Wu, Shengqiong and Zheng, Yanhao and Ji, Jiayi and Zhou, Fan and Luo, Jiebo and Liu, Ziwei and Fei, Hao and Chua, Tat-Seng},
conference = {The Fourteenth International Conference on Learning Representations},
year = {2026},
}