BIT’s system for AutoSimulTrans2021

Abstract

In this paper we introduce our Chinese-English simultaneous translation system participating in AutoSimTrans 2021. In simultaneous translation, translation quality and latency are both important. In order to reduce the translation latency, we cut the streaming-input source sentence into segments and translate the segments before the full sentence is received. In order to obtain high-quality translations, we pretrain a translation model with adequate corpus and fine-tune the model with domain adaptation and sentence length adaptation. The experimental results on the development dataset show that our system performs better than the baseline system.