1 Introduction

1.1 Problem Statement

learn2listen

1.2 Stated Contribution

2 Related Work

2.1 Conversational dynamics

2.2 Text-conditioned motion synthesis

3 Method

3.1 Problem Definition

3.2 Discretizing Listener Motion

VQVAE学习一个motion表征

3.3 Text-Conditioned Motion Generation

用LM做Generator，输入VQ Token

4 Experiment

4.1 Implementation Details

5 Summary

5.1 Relative Position

6 Comments