NetEase-GameAI / ChoreoMaster

This repository contains the dataset used in paper "ChoreoMaster: Choreography -Oriented Music Driven Dance Synthesis".
112 stars 10 forks source link

"ChoreoMaster: Choreography -Oriented Music Driven Dance Synthesis"

1. Description

To facilitate further research in this area, we release the copyright-free portion of our paired music/motion dataset (1.48 hours out of 9.91 hours). It contains 2708 dance motion meters (the motion unit used to construct motion graph), all paired with music clips and labels, with a total frame count of 160136 sampled at 30 fps.

The motion, music and label files are respectively stored in three folders ("label_json", "motion_fbx", "music_clip") with the same prefix name. All the motion files are saved as .fbx format and we also provide a T-pos model file.

This dataset can only be used for the sole purpose of performing non-commercial scientific research. Any other use, in particular any use for commercial purposes, is prohibited.

2. Label Format:

The label file are saved as json format, containing the two dimentional music and motion styles and the corresponding rhythm signature.

{
    "name": the motion clip name with the format of motion_id@clip_id,  consecutive clips will have the same motion_id with consecutive clip_id
    "music_style1": music style 1,
    "music_style2": music style 2,
    "motion_style1": motion style 1,
    "motion_style2": motion style 2,
    "signature": 8 bit rhythm signature string
}

Details about the style labels and rhythm signatures can be found in our paper.

3. Publications:

If you use our dataset in your work, please cite our article:

@article{choreomaster2021,
    author = {Chen Kang, Zhipeng Tan, Jin Lei, Song-Hai Zhang, Yuan-Chen Guo, Weidong Zhang, Shi-Min Hu},
    title = {ChoreoMaster : Choreography-Oriented Music-Driven Dance Synthesis},
    journal = {ACM Transactions on Graphics (TOG)},
    volume = {40},
    number = {4},
    year = {2021},
    publisher = {ACM}
  }