We present a method that gets as input an audio of violin or piano playing, and outputs a video of skeleton predictions which are further used to animate an avatar. The key idea is to create an animation of an avatar that moves their hands similarly to how a pianist or violinist would do, just from audio. Aiming for a fully detailed correct arms and fingers motion is the ultimate goal, however, it's not clear if body movement can be predicted from music at all. In this paper, we present the first result that shows that natural body dynamics can be predicted. We built an LSTM network that is trained on violin and piano recital videos uploaded to the Internet. The predicted points are applied onto a rigged avatar to create the animation.




Paper pdf; Published in CVPR 2018 (Spotlight).

  title={Audio to Body Dynamics},
  author={Shlizerman, Eli and Dery, Lucio M and Schoen, Hayden and Kemelmacher-Shlizerman, Ira},
  journal={CVPR, IEEE Computer Society Conference on Computer Vision and Pattern Recognition},


Eli Shlizerman shlizee@fb.com, Ira Kemelmacher-Shlizerman kemelmi@fb.com