Streaming automatic speech recognition (ASR) must be judged jointly on what it transcribes and on how quickly it commits each word. Delayed streams modeling …
机构:UIUC
来源:arXiv 2609.18333 | AI4Papers 论文推荐平台