[2605.31432] DOA: Training-Free Decoder-Only Attention Policy for Long-Form Simultaneous Translation with SpeechLLMs
![[2605.31432] DOA: Training-Free Decoder-Only Attention Policy for Long-Form Simultaneous Translation with SpeechLLMs](/static/img/curated/afbytes-us.png)
AI disclosure
Summary
Abstract page for arXiv paper 2605.31432: DOA: Training-Free Decoder-Only Attention Policy for Long-Form Simultaneous Translation with SpeechLLMs