新闻
AURA: AUdio-dRiven streaming Avatar
arXiv cs.AI · 发布于 · 阅读约3分钟
30秒读懂
- 发生了什么
- AURA generates streaming talking-head video from audio in real time on single GPUs, requiring only current audio chunk without multi-second lookahead.
- 为何重要
- Matters for engineers building voice assistants, virtual avatars, and interactive applications needing low-latency video generation from speech.
- 注意
- Smaller model streams 576p at 20.7 fps; larger model streams 720p at 9.7 fps. Quality tradeoffs exist compared to non-streaming bidirectional models.
- latency
- kv cache
- voice
这条新闻背后的模式
每个模式都讲清楚技术如何运作、何时值得投入,以及在哪里会失效。
The Agent Architect
每周一个模式、一个权衡、一个生产事故案例。为构建智能体系统的人准备的每周简报。
每周一封邮件,一键退订。您的地址仅用于发送简报。