@inproceedings{zanon-boito-etal-2022-trac,
title = "{ON}-{TRAC} Consortium Systems for the {IWSLT} 2022 Dialect and Low-resource Speech Translation Tasks",
author = {Zanon Boito, Marcely and
Ortega, John and
Riguidel, Hugo and
Laurent, Antoine and
Barrault, Lo{\"\i}c and
Bougares, Fethi and
Chaabani, Firas and
Nguyen, Ha and
Barbier, Florentin and
Gahbiche, Souhir and
Est{\`e}ve, Yannick},
editor = "Salesky, Elizabeth and
Federico, Marcello and
Costa-juss{\`a}, Marta",
booktitle = "Proceedings of the 19th International Conference on Spoken Language Translation (IWSLT 2022)",
month = may,
year = "2022",
address = "Dublin, Ireland (in-person and online)",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2022.iwslt-1.28",
doi = "10.18653/v1/2022.iwslt-1.28",
pages = "308--318",
abstract = "This paper describes the ON-TRAC Consortium translation systems developed for two challenge tracks featured in the Evaluation Campaign of IWSLT 2022: low-resource and dialect speech translation. For the Tunisian Arabic-English dataset (low-resource and dialect tracks), we build an end-to-end model as our joint primary submission, and compare it against cascaded models that leverage a large fine-tuned wav2vec 2.0 model for ASR. Our results show that in our settings pipeline approaches are still very competitive, and that with the use of transfer learning, they can outperform end-to-end models for speech translation (ST). For the Tamasheq-French dataset (low-resource track) our primary submission leverages intermediate representations from a wav2vec 2.0 model trained on 234 hours of Tamasheq audio, while our contrastive model uses a French phonetic transcription of the Tamasheq audio as input in a Conformer speech translation architecture jointly trained on automatic speech recognition, ST and machine translation losses. Our results highlight that self-supervised models trained on smaller sets of target data are more effective to low-resource end-to-end ST fine-tuning, compared to large off-the-shelf models. Results also illustrate that even approximate phonetic transcriptions can improve ST scores.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="zanon-boito-etal-2022-trac">
<titleInfo>
<title>ON-TRAC Consortium Systems for the IWSLT 2022 Dialect and Low-resource Speech Translation Tasks</title>
</titleInfo>
<name type="personal">
<namePart type="given">Marcely</namePart>
<namePart type="family">Zanon Boito</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">John</namePart>
<namePart type="family">Ortega</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hugo</namePart>
<namePart type="family">Riguidel</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Antoine</namePart>
<namePart type="family">Laurent</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Loïc</namePart>
<namePart type="family">Barrault</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Fethi</namePart>
<namePart type="family">Bougares</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Firas</namePart>
<namePart type="family">Chaabani</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ha</namePart>
<namePart type="family">Nguyen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Florentin</namePart>
<namePart type="family">Barbier</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Souhir</namePart>
<namePart type="family">Gahbiche</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yannick</namePart>
<namePart type="family">Estève</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2022-05</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 19th International Conference on Spoken Language Translation (IWSLT 2022)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Elizabeth</namePart>
<namePart type="family">Salesky</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Marcello</namePart>
<namePart type="family">Federico</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Marta</namePart>
<namePart type="family">Costa-jussà</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Dublin, Ireland (in-person and online)</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>This paper describes the ON-TRAC Consortium translation systems developed for two challenge tracks featured in the Evaluation Campaign of IWSLT 2022: low-resource and dialect speech translation. For the Tunisian Arabic-English dataset (low-resource and dialect tracks), we build an end-to-end model as our joint primary submission, and compare it against cascaded models that leverage a large fine-tuned wav2vec 2.0 model for ASR. Our results show that in our settings pipeline approaches are still very competitive, and that with the use of transfer learning, they can outperform end-to-end models for speech translation (ST). For the Tamasheq-French dataset (low-resource track) our primary submission leverages intermediate representations from a wav2vec 2.0 model trained on 234 hours of Tamasheq audio, while our contrastive model uses a French phonetic transcription of the Tamasheq audio as input in a Conformer speech translation architecture jointly trained on automatic speech recognition, ST and machine translation losses. Our results highlight that self-supervised models trained on smaller sets of target data are more effective to low-resource end-to-end ST fine-tuning, compared to large off-the-shelf models. Results also illustrate that even approximate phonetic transcriptions can improve ST scores.</abstract>
<identifier type="citekey">zanon-boito-etal-2022-trac</identifier>
<identifier type="doi">10.18653/v1/2022.iwslt-1.28</identifier>
<location>
<url>https://aclanthology.org/2022.iwslt-1.28</url>
</location>
<part>
<date>2022-05</date>
<extent unit="page">
<start>308</start>
<end>318</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T ON-TRAC Consortium Systems for the IWSLT 2022 Dialect and Low-resource Speech Translation Tasks
%A Zanon Boito, Marcely
%A Ortega, John
%A Riguidel, Hugo
%A Laurent, Antoine
%A Barrault, Loïc
%A Bougares, Fethi
%A Chaabani, Firas
%A Nguyen, Ha
%A Barbier, Florentin
%A Gahbiche, Souhir
%A Estève, Yannick
%Y Salesky, Elizabeth
%Y Federico, Marcello
%Y Costa-jussà, Marta
%S Proceedings of the 19th International Conference on Spoken Language Translation (IWSLT 2022)
%D 2022
%8 May
%I Association for Computational Linguistics
%C Dublin, Ireland (in-person and online)
%F zanon-boito-etal-2022-trac
%X This paper describes the ON-TRAC Consortium translation systems developed for two challenge tracks featured in the Evaluation Campaign of IWSLT 2022: low-resource and dialect speech translation. For the Tunisian Arabic-English dataset (low-resource and dialect tracks), we build an end-to-end model as our joint primary submission, and compare it against cascaded models that leverage a large fine-tuned wav2vec 2.0 model for ASR. Our results show that in our settings pipeline approaches are still very competitive, and that with the use of transfer learning, they can outperform end-to-end models for speech translation (ST). For the Tamasheq-French dataset (low-resource track) our primary submission leverages intermediate representations from a wav2vec 2.0 model trained on 234 hours of Tamasheq audio, while our contrastive model uses a French phonetic transcription of the Tamasheq audio as input in a Conformer speech translation architecture jointly trained on automatic speech recognition, ST and machine translation losses. Our results highlight that self-supervised models trained on smaller sets of target data are more effective to low-resource end-to-end ST fine-tuning, compared to large off-the-shelf models. Results also illustrate that even approximate phonetic transcriptions can improve ST scores.
%R 10.18653/v1/2022.iwslt-1.28
%U https://aclanthology.org/2022.iwslt-1.28
%U https://doi.org/10.18653/v1/2022.iwslt-1.28
%P 308-318
Markdown (Informal)
[ON-TRAC Consortium Systems for the IWSLT 2022 Dialect and Low-resource Speech Translation Tasks](https://aclanthology.org/2022.iwslt-1.28) (Zanon Boito et al., IWSLT 2022)
ACL
- Marcely Zanon Boito, John Ortega, Hugo Riguidel, Antoine Laurent, Loïc Barrault, Fethi Bougares, Firas Chaabani, Ha Nguyen, Florentin Barbier, Souhir Gahbiche, and Yannick Estève. 2022. ON-TRAC Consortium Systems for the IWSLT 2022 Dialect and Low-resource Speech Translation Tasks. In Proceedings of the 19th International Conference on Spoken Language Translation (IWSLT 2022), pages 308–318, Dublin, Ireland (in-person and online). Association for Computational Linguistics.