Audio large language models (AudioLLMs) operate reactively, responding only when queried. We introduce proactive audio assistance, where an AudioLLM monitors…
机构:Meta
来源:arXiv 2609.21183 | AI4Papers 论文推荐平台