Towards Efficient Simultaneous Speech Translation: CUNI-KIT System for Simultaneous Track at IWSLT 2023

Peter Polák, Danni Liu, Ngoc-Quan Pham, Jan Niehues, Alexander Waibel, Ondřej Bojar

Abstract
In this paper, we describe our submission to the Simultaneous Track at IWSLT 2023. This year, we continue with the successful setup from the last year, however, we adopt the latest methods that further improve the translation quality. Additionally, we propose a novel online policy for attentional encoder-decoder models. The policy prevents the model to generate translation beyond the current speech input by using an auxiliary CTC output layer. We show that the proposed simultaneous policy can be applied to both streaming blockwise models and offline encoder-decoder models. We observe significant improvements in quality (up to 1.1 BLEU) and the computational footprint (up to 45% relative RTF).
Anthology ID:
2023.iwslt-1.37
Volume:
Proceedings of the 20th International Conference on Spoken Language Translation (IWSLT 2023)
Month:
July
Year:
2023
Address:
Toronto, Canada (in-person and online)
Editors:
Elizabeth Salesky, Marcello Federico, Marine Carpuat
Venue:
IWSLT
SIG:
SIGSLT
Publisher:
Association for Computational Linguistics
Note:
Pages:
389–396
Language:
URL:
https://preview.aclanthology.org/icon-24-ingestion/2023.iwslt-1.37/
DOI:
10.18653/v1/2023.iwslt-1.37
Bibkey:
Cite (ACL):
Peter Polák, Danni Liu, Ngoc-Quan Pham, Jan Niehues, Alexander Waibel, and Ondřej Bojar. 2023. Towards Efficient Simultaneous Speech Translation: CUNI-KIT System for Simultaneous Track at IWSLT 2023. In Proceedings of the 20th International Conference on Spoken Language Translation (IWSLT 2023), pages 389–396, Toronto, Canada (in-person and online). Association for Computational Linguistics.
Cite (Informal):
Towards Efficient Simultaneous Speech Translation: CUNI-KIT System for Simultaneous Track at IWSLT 2023 (Polák et al., IWSLT 2023)
Copy Citation:
PDF:
https://preview.aclanthology.org/icon-24-ingestion/2023.iwslt-1.37.pdf
PDF Search Fix data