Speech large language models (SpeechLLMs) offer reduced latency and retain paralinguistic nuances that are typically lost in cascaded automatic speech recogn…
机构:UT Austin
来源:arXiv 2609.11864 | AI4Papers 论文推荐平台