Zenodo.org will be unavailable for 2 hours on September 29th from 06:00-08:00 UTC. See announcement.

Conference paper Closed Access

ESPnet How2 Speech Translation System for IWSLT 2019: Pre-training, Knowledge Distillation, and Going Deeper

Inaguma, Hirofumi; Kiyono, Shun; Soplin, Nelson Enrique Yalta; Suzuki, Jun; Duh, Kevin; Watanabe, Shinji

This paper describes the ESPnet submissions to the How2 Speech Translation task at IWSLT2019. In this year, we mainly build our systems based on Transformer architectures in all tasks and focus on the end-to-end speech translation (E2E-ST). We first compare RNN-based models and Transformer, and then confirm Transformer models significantly and consistently outperform RNN models in all tasks and corpora. Next, we investigate pre-training of E2E-ST models with the ASR and MT tasks. On top of the pre-training, we further explore knowledge distillation from the NMT model and the deeper speech encoder, and confirm drastic improvements over the baseline model. All of our codes are publicly available in ESPnet.

Closed Access

Files are not publicly accessible.

All versions This version
Views 338338
Downloads 2222
Data volume 3.8 MB3.8 MB
Unique views 288288
Unique downloads 1919


Cite as