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

Sara Papi, Marco Turchi, Matteo Negri

Browse the full Interspeech paper archive.