Recent advances in world models have enabled increasingly realistic visual synthesis. However, these generated environments remain largely silent. Bringing sound to world models poses three core challenges: real-time generation to keep pace with interactive video streams, interactive control to respond to mid-stream sound instructions, and spatially aligned stereo to reflect scene geometry and cam

---
**📖 中文解读**
以上内容由AI翻译自英文原文,可能存在不准确之处。建议阅读[原文](https://arxiv.org/abs/2610.08760v1)获取最准确的信息。

---
🔗 **原文链接**: [WorldSonus: Bringing Sound to Worlds](https://arxiv.org/abs/2610.08760v1)
🏷️ **转载来源**: ArXiv cs.AI
> 本文由小九AI技术站翻译整理,内容版权归原作者所有。
👤 作者: Pengjun Fang, Jingyi Fa, Kam Man Wu, Jiaming Wang, Haoyuan Huang, Yaguang Wu, Xiangjun Huang, 马子阳, Weijia Chen, Hongyu Liu, Zeyue Tian, Qifeng Chen

---
🐾 **小九锐评**

这篇论文来自arXiv预印本,虽然还没有经过同行评审,但选题方向值得关注。
建议先读中文摘要判断是否相关,再看全文细节。

你对这个话题有什么看法?欢迎在评论区讨论 💬

> _转载自 ArXiv cs.AI,内容版权归原作者所有_

---
⏱️ 2026-10-07 13:03