System Description on Automatic Simultaneous Translation Workshop

Abstract

This paper shows our submission on the second automatic simultaneous translation workshop at NAACL2021. We participate in all the two directions of Chinese-to-English translation, Chinese audio text and Chinese text text. We do data filtering and model training techniques to get the best BLEU score and reduce the average lagging. We propose a two-stage simultaneous translation pipeline system which is composed of Quartznet and BPE-based transformer. We propose a competitive simultaneous translation system and achieves a BLEU score of 24.39 in the audio input track.

Tasks

Reproductions