AlignAtt: Using Attention-based Audio-Translation Alignments as a Guide for Simultaneous Speech Translation

Open in new window