Wan-Streamer v0.1: End-to-end Real-time Interactive Foundation Models

Arxiv CS.SD (Sound/Audio ML)

Wan-Streamer is a native-streaming, full-duplex interactive model that jointly processes language, audio, and video in real time. The v0.1 release includes a project page and HuggingFace demo.

wan-streamer

See 1 more source

Wan-Streamer v0.1: End-to-end Real-time Interactive Foundation Models4 hours agoGraceToSentienceDiscuss

Arxiv CS.SD (Sound/Audio ML)·Lianghua HuangZhifan WuWei WangYupeng ShiMengyang FengJunjie HeChenwei XieYu LiuJingren ZhouAng WangBang ZhangBaole AiChen LiangCheng YuChongyang ZhongJinwei QiKai ZhuPandeng LiPeng ZhangWenyuan ZhangXinhua ChengYi

Jun 25, 4:00 AM