YGD-mix / README.md
alibabasglab's picture
Update README.md
b740ad2 verified
metadata
license: bsd-3-clause
tags:
  - speaker extraction
  - speech separation
  - gesture
  - pose
  - multimodal
  - cocktail party problem

A modified version of the Youtube Gesture Dataset. Original data can be downloaded and processed here and here.

This dataset is used for Audio-visual speaker extraction conditioned on body gestures in the SEG paper, which the code can be found here.