AI RESEARCH
EchoAvatar: Real-time Generative Avatar Animation from Audio Streams
arXiv CS.CV
•
ArXi:2605.28272v1 Announce Type: new Real-time synthesis of high-fidelity 3D character motion from audio is a pivotal component for next-generation interactive avatars and virtual assistants. However, most existing approaches are limited to offline processing of complete audio sequences or are constrained to specific domains, rarely handling both speech and music effectively. In this paper, we