Adapting ASR to a domain with only untranscribed audio typically relies on pseudo-labeling, a strong teacher transcribes the audio and a student is fine-tune…
机构:Idiap
来源:arXiv 2610.32572 | AI4Papers 论文推荐平台