@inproceedings{wu-etal-2024-improving-quality,
title = "Improving the Quality of {IWLST} 2024 Cascade Offline Speech Translation and Speech-to-Speech Translation via Translation Hypothesis Ensembling with {NMT} models and Large Language Models",
author = "Wu, Zhanglin and
Guo, Jiaxin and
Wei, Daimeng and
Rao, Zhiqiang and
Li, Zongyao and
Shang, Hengchao and
Luo, Yuanchang and
Li, Shaojun and
Yang, Hao",
editor = "Salesky, Elizabeth and
Federico, Marcello and
Carpuat, Marine",
booktitle = "Proceedings of the 21st International Conference on Spoken Language Translation (IWSLT 2024)",
month = aug,
year = "2024",
address = "Bangkok, Thailand (in-person and online)",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.iwslt-1.7",
pages = "46--52",
abstract = "This paper presents HW-TSC{'}s submission to the IWSLT 2024 Offline Speech Translation Task and Speech-to-Speech Translation Task. The former includes three translation directions: English to German, English to Chinese, and English to Japanese, while the latter only includes the translation direction of English to Chinese. We attend all three tracks (Constraint training, Constrained with Large Language Models training, and Unconstrained training) of offline speech translation task, using the cascade model architecture. Under the constrained training track, we train an ASR model from scratch, and then employ R-Drop and domain data selection to train the NMT model. In the constrained with Large Language Models training track, we use Wav2vec 2.0 and mBART50 for ASR model training initialization, and then train the LLama2-7B-based MT model using continuous training with sentence-aligned parallel data, supervised fine-tuning, and contrastive preference optimization. In the unconstrained training track, we fine-tune the whisper model for speech recognition, and then ensemble the translation results of NMT models and LLMs to produce superior translation output. For the speech-to-speech translation Task, we initially employ the offline speech translation system described above to generate the translated text. Then, we utilize the VITS model to generate the corresponding speech and employ the OpenVoice model for timbre cloning.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="wu-etal-2024-improving-quality">
<titleInfo>
<title>Improving the Quality of IWLST 2024 Cascade Offline Speech Translation and Speech-to-Speech Translation via Translation Hypothesis Ensembling with NMT models and Large Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Zhanglin</namePart>
<namePart type="family">Wu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiaxin</namePart>
<namePart type="family">Guo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Daimeng</namePart>
<namePart type="family">Wei</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhiqiang</namePart>
<namePart type="family">Rao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zongyao</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hengchao</namePart>
<namePart type="family">Shang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yuanchang</namePart>
<namePart type="family">Luo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shaojun</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hao</namePart>
<namePart type="family">Yang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2024-08</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 21st International Conference on Spoken Language Translation (IWSLT 2024)</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">Marine</namePart>
<namePart type="family">Carpuat</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Bangkok, Thailand (in-person and online)</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>This paper presents HW-TSC’s submission to the IWSLT 2024 Offline Speech Translation Task and Speech-to-Speech Translation Task. The former includes three translation directions: English to German, English to Chinese, and English to Japanese, while the latter only includes the translation direction of English to Chinese. We attend all three tracks (Constraint training, Constrained with Large Language Models training, and Unconstrained training) of offline speech translation task, using the cascade model architecture. Under the constrained training track, we train an ASR model from scratch, and then employ R-Drop and domain data selection to train the NMT model. In the constrained with Large Language Models training track, we use Wav2vec 2.0 and mBART50 for ASR model training initialization, and then train the LLama2-7B-based MT model using continuous training with sentence-aligned parallel data, supervised fine-tuning, and contrastive preference optimization. In the unconstrained training track, we fine-tune the whisper model for speech recognition, and then ensemble the translation results of NMT models and LLMs to produce superior translation output. For the speech-to-speech translation Task, we initially employ the offline speech translation system described above to generate the translated text. Then, we utilize the VITS model to generate the corresponding speech and employ the OpenVoice model for timbre cloning.</abstract>
<identifier type="citekey">wu-etal-2024-improving-quality</identifier>
<location>
<url>https://aclanthology.org/2024.iwslt-1.7</url>
</location>
<part>
<date>2024-08</date>
<extent unit="page">
<start>46</start>
<end>52</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Improving the Quality of IWLST 2024 Cascade Offline Speech Translation and Speech-to-Speech Translation via Translation Hypothesis Ensembling with NMT models and Large Language Models
%A Wu, Zhanglin
%A Guo, Jiaxin
%A Wei, Daimeng
%A Rao, Zhiqiang
%A Li, Zongyao
%A Shang, Hengchao
%A Luo, Yuanchang
%A Li, Shaojun
%A Yang, Hao
%Y Salesky, Elizabeth
%Y Federico, Marcello
%Y Carpuat, Marine
%S Proceedings of the 21st International Conference on Spoken Language Translation (IWSLT 2024)
%D 2024
%8 August
%I Association for Computational Linguistics
%C Bangkok, Thailand (in-person and online)
%F wu-etal-2024-improving-quality
%X This paper presents HW-TSC’s submission to the IWSLT 2024 Offline Speech Translation Task and Speech-to-Speech Translation Task. The former includes three translation directions: English to German, English to Chinese, and English to Japanese, while the latter only includes the translation direction of English to Chinese. We attend all three tracks (Constraint training, Constrained with Large Language Models training, and Unconstrained training) of offline speech translation task, using the cascade model architecture. Under the constrained training track, we train an ASR model from scratch, and then employ R-Drop and domain data selection to train the NMT model. In the constrained with Large Language Models training track, we use Wav2vec 2.0 and mBART50 for ASR model training initialization, and then train the LLama2-7B-based MT model using continuous training with sentence-aligned parallel data, supervised fine-tuning, and contrastive preference optimization. In the unconstrained training track, we fine-tune the whisper model for speech recognition, and then ensemble the translation results of NMT models and LLMs to produce superior translation output. For the speech-to-speech translation Task, we initially employ the offline speech translation system described above to generate the translated text. Then, we utilize the VITS model to generate the corresponding speech and employ the OpenVoice model for timbre cloning.
%U https://aclanthology.org/2024.iwslt-1.7
%P 46-52
Markdown (Informal)
[Improving the Quality of IWLST 2024 Cascade Offline Speech Translation and Speech-to-Speech Translation via Translation Hypothesis Ensembling with NMT models and Large Language Models](https://aclanthology.org/2024.iwslt-1.7) (Wu et al., IWSLT 2024)
ACL
- Zhanglin Wu, Jiaxin Guo, Daimeng Wei, Zhiqiang Rao, Zongyao Li, Hengchao Shang, Yuanchang Luo, Shaojun Li, and Hao Yang. 2024. Improving the Quality of IWLST 2024 Cascade Offline Speech Translation and Speech-to-Speech Translation via Translation Hypothesis Ensembling with NMT models and Large Language Models. In Proceedings of the 21st International Conference on Spoken Language Translation (IWSLT 2024), pages 46–52, Bangkok, Thailand (in-person and online). Association for Computational Linguistics.