ON-TRAC Consortium for End-to-End and Simultaneous Speech Translation Challenge Tasks at IWSLT 2020

This paper describes the ON-TRAC Consortium translation systems developed for\ntwo challenge tracks featured in the Evaluation Campaign of IWSLT 2020, offline\nspeech translation and simultaneous speech translation. ON-TRAC Consortium is\ncomposed of researchers from three French academic laboratories: LIA (Avignon\nUniversit\\'e), LIG (Universit\\'e Grenoble Alpes), and LIUM (Le Mans\nUniversit\\'e). Attention-based encoder-decoder models, trained end-to-end, were\nused for our submissions to the offline speech translation track. Our\ncontributions focused on data augmentation and ensembling of multiple models.\nIn the simultaneous speech translation track, we build on Transformer-based\nwait-k models for the text-to-text subtask. For speech-to-text simultaneous\ntranslation, we attach a wait-k MT system to a hybrid ASR system. We propose an\nalgorithm to control the latency of the ASR+MT cascade and achieve a good\nlatency-quality trade-off on both subtasks.\n