CUNI-KIT System for Simultaneous Speech Translation Task at IWSLT 2022

P. Polák; N.Q. Pham; T.N. Nguyen; D.N. Liu; C. Mullov; J. Niehues; O. Bojar; A. Waibel

CUNI-KIT System for Simultaneous Speech Translation Task at IWSLT 2022

P. Polák^*, N.Q. Pham, T.N. Nguyen, D.N. Liu, C. Mullov, J. Niehues, O. Bojar, A. Waibel

^*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceeding › Conference article in proceeding › Academic › peer-review

Abstract

In this paper, we describe our submission to the Simultaneous Speech Translation at IWSLT 2022. We explore strategies to utilize an offline model in a simultaneous setting without the need to modify the original model. In our experiments, we show that our onlinization algorithm is almost on par with the offline setting while being 3x faster than offline in terms of latency on the test set. We also show that the onlinized offline model outperforms the best IWSLT2021 simultaneous system in medium and high latency regimes and is almost on par in the low latency regime. We make our system publicly available.

Original language	English
Title of host publication	Proceedings of the 19th International Conference on Spoken Language Translation (IWSLT 2022)
Publisher	Association for Computational Linguistics
Pages	277-285
Number of pages	9
ISBN (Print)	9781955917414
Publication status	Published - May 2022
Event	The International Conference on Spoken Language Translation - Dublin, Ireland Duration: 26 May 2022 → 27 May 2022 https://iwslt.org/2022/

Conference

Conference	The International Conference on Spoken Language Translation
Abbreviated title	19th IWSLT
Country/Territory	Ireland
City	Dublin
Period	26/05/22 → 27/05/22
Internet address	https://iwslt.org/2022/

Access to Document

https://aclanthology.org/2022.iwslt-1.24.pdf

Cite this

Polák, P., Pham, N. Q., Nguyen, T. N., Liu, D. N., Mullov, C., Niehues, J., Bojar, O., & Waibel, A. (2022). CUNI-KIT System for Simultaneous Speech Translation Task at IWSLT 2022. In Proceedings of the 19th International Conference on Spoken Language Translation (IWSLT 2022) (pp. 277-285). Association for Computational Linguistics. https://aclanthology.org/2022.iwslt-1.24.pdf

@inproceedings{833884c1e382468da16b70c40eb2f4c5,

title = "CUNI-KIT System for Simultaneous Speech Translation Task at IWSLT 2022",

abstract = "In this paper, we describe our submission to the Simultaneous Speech Translation at IWSLT 2022. We explore strategies to utilize an offline model in a simultaneous setting without the need to modify the original model. In our experiments, we show that our onlinization algorithm is almost on par with the offline setting while being 3x faster than offline in terms of latency on the test set. We also show that the onlinized offline model outperforms the best IWSLT2021 simultaneous system in medium and high latency regimes and is almost on par in the low latency regime. We make our system publicly available.",

author = "P. Pol{\'a}k and N.Q. Pham and T.N. Nguyen and D.N. Liu and C. Mullov and J. Niehues and O. Bojar and A. Waibel",

note = "Funding Information: This work has received support from the project “Grant Schemes at CU” (reg. no. CZ.02.2.69/0.0/0.0/19_073/0016935), the grant 19-26934X (NEUREM3) of the Czech Science Foundation, the European Union{\textquoteright}s Horizon 2020 Research and Innovation Programme under Grant Agreement No 825460 (ELITR), and partly supported by a Facebook Sponsored Research Agreement “Language Similarity in Machine Translation”. Publisher Copyright: {\textcopyright} 2022 Association for Computational Linguistics.; The International Conference on Spoken Language Translation, 19th IWSLT ; Conference date: 26-05-2022 Through 27-05-2022",

year = "2022",

month = may,

language = "English",

isbn = "9781955917414",

pages = "277--285",

booktitle = "Proceedings of the 19th International Conference on Spoken Language Translation (IWSLT 2022)",

publisher = "Association for Computational Linguistics",

url = "https://iwslt.org/2022/",

}

Polák, P, Pham, NQ, Nguyen, TN, Liu, DN, Mullov, C, Niehues, J, Bojar, O & Waibel, A 2022, CUNI-KIT System for Simultaneous Speech Translation Task at IWSLT 2022. in Proceedings of the 19th International Conference on Spoken Language Translation (IWSLT 2022). Association for Computational Linguistics, pp. 277-285, The International Conference on Spoken Language Translation, Dublin, Ireland, 26/05/22. <https://aclanthology.org/2022.iwslt-1.24.pdf>

CUNI-KIT System for Simultaneous Speech Translation Task at IWSLT 2022. / Polák, P.; Pham, N.Q.; Nguyen, T.N. et al.
Proceedings of the 19th International Conference on Spoken Language Translation (IWSLT 2022). Association for Computational Linguistics, 2022. p. 277-285.

Research output: Chapter in Book/Report/Conference proceeding › Conference article in proceeding › Academic › peer-review

TY - GEN

T1 - CUNI-KIT System for Simultaneous Speech Translation Task at IWSLT 2022

AU - Polák, P.

AU - Pham, N.Q.

AU - Nguyen, T.N.

AU - Liu, D.N.

AU - Mullov, C.

AU - Niehues, J.

AU - Bojar, O.

AU - Waibel, A.

N1 - Funding Information: This work has received support from the project “Grant Schemes at CU” (reg. no. CZ.02.2.69/0.0/0.0/19_073/0016935), the grant 19-26934X (NEUREM3) of the Czech Science Foundation, the European Union’s Horizon 2020 Research and Innovation Programme under Grant Agreement No 825460 (ELITR), and partly supported by a Facebook Sponsored Research Agreement “Language Similarity in Machine Translation”. Publisher Copyright: © 2022 Association for Computational Linguistics.

PY - 2022/5

Y1 - 2022/5

N2 - In this paper, we describe our submission to the Simultaneous Speech Translation at IWSLT 2022. We explore strategies to utilize an offline model in a simultaneous setting without the need to modify the original model. In our experiments, we show that our onlinization algorithm is almost on par with the offline setting while being 3x faster than offline in terms of latency on the test set. We also show that the onlinized offline model outperforms the best IWSLT2021 simultaneous system in medium and high latency regimes and is almost on par in the low latency regime. We make our system publicly available.

AB - In this paper, we describe our submission to the Simultaneous Speech Translation at IWSLT 2022. We explore strategies to utilize an offline model in a simultaneous setting without the need to modify the original model. In our experiments, we show that our onlinization algorithm is almost on par with the offline setting while being 3x faster than offline in terms of latency on the test set. We also show that the onlinized offline model outperforms the best IWSLT2021 simultaneous system in medium and high latency regimes and is almost on par in the low latency regime. We make our system publicly available.

M3 - Conference article in proceeding

SN - 9781955917414

SP - 277

EP - 285

BT - Proceedings of the 19th International Conference on Spoken Language Translation (IWSLT 2022)

PB - Association for Computational Linguistics

T2 - The International Conference on Spoken Language Translation

Y2 - 26 May 2022 through 27 May 2022

ER -