Reproduce “ECHO-G: Embodied Co-speech Humanoid mOtion Generation” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Yizhao Li, Pusen Gao, Ming Wang, Shaojie Shen, Shuo Yang, Hao Xu

The ECHO-G system generates executable humanoid robot-motion references directly from speech audio and a transcript with word timings. Audio guides movement timing, while timed words provide linguistic context. On speaker-held-out evaluation, it lowers Fréchet Gesture Distance to 2.278 from 4.725 and processing time to 5.96 from 6.56 milliseconds per frame, but not every physical-quality measure.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.