TRCA: Text Restoration for Chinese ASR with BERT

Xing Wu; Yuan Zhang; Jianjia Wang; Yike Guo

doi:10.3233/FAIA220295

TRCA: Text Restoration for Chinese ASR with BERT

Xing Wu^*
, Yuan Zhang
, Jianjia Wang
, Yike Guo

^*Corresponding author for this work

Office of the Vice-President (Research and Development)

Research output: Chapter in book/report/conference proceeding › Conference proceeding › peer-review

Abstract

Text restoration plays a vital role in Chinese automatic speech recognition (ASR), which includes punctuation prediction and error correction. However, there are two inevitable challenges for this task. On the one hand, there are no public dataset and model for Chinese punctuation prediction. On the other hand, current text restoration methods for automatic speech recognition only focus on Chinese error correction instead of combining with Chinese punctuation prediction task. To address these problems, a BERT-based text restoration method called TRCA is proposed for Chinese ASR consisting of a Chinese punctuation prediction model and a Chinese error correction model. Experiments demonstrate that the proposed TRCA method outperforms state-of-the-art methods for both punctuation prediction and error correction tasks, among which the proposed TRCA improves the average accuracy to 98% in Chinese punctuation prediction.

Original language	English
Title of host publication	New Trends in Intelligent Software Methodologies, Tools and Techniques - Proceedings of the 21st International Conference on New Trends in Intelligent Software Methodologies, Tools and Techniques, SoMeT 2022
Editors	Hamido Fujita, Yutaka Watanobe, Takuya Azumi
Publisher	IOS Press BV
Pages	661-668
Number of pages	8
ISBN (Electronic)	9781643683171
ISBN (Print)	9781643683164
DOIs	https://doi.org/10.3233/FAIA220295
Publication status	Published - 14 Sept 2022
Event	21st International Conference on New Trends in Intelligent Software Methodologies, Tools and Techniques, SoMeT 2022 - Kitakyushu, Japan Duration: 20 Sept 2022 → 22 Sept 2022

Publication series

Name	Frontiers in Artificial Intelligence and Applications
Volume	355
ISSN (Print)	0922-6389
ISSN (Electronic)	1879-8314

Conference

Conference	21st International Conference on New Trends in Intelligent Software Methodologies, Tools and Techniques, SoMeT 2022
Country/Territory	Japan
City	Kitakyushu
Period	20/09/22 → 22/09/22

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 9 Industry, Innovation, and Infrastructure

User-Defined Keywords

BERT
Chinese ASR text restoration
Chinese error correction
Chinese punctuation prediction

Access to Document

10.3233/FAIA220295

Cite this

Wu, X., Zhang, Y., Wang, J., & Guo, Y. (2022). TRCA: Text Restoration for Chinese ASR with BERT. In H. Fujita, Y. Watanobe, & T. Azumi (Eds.), New Trends in Intelligent Software Methodologies, Tools and Techniques - Proceedings of the 21st International Conference on New Trends in Intelligent Software Methodologies, Tools and Techniques, SoMeT 2022 (pp. 661-668). (Frontiers in Artificial Intelligence and Applications; Vol. 355). IOS Press BV. https://doi.org/10.3233/FAIA220295

Wu, Xing ; Zhang, Yuan ; Wang, Jianjia et al. / TRCA : Text Restoration for Chinese ASR with BERT. New Trends in Intelligent Software Methodologies, Tools and Techniques - Proceedings of the 21st International Conference on New Trends in Intelligent Software Methodologies, Tools and Techniques, SoMeT 2022. editor / Hamido Fujita ; Yutaka Watanobe ; Takuya Azumi. IOS Press BV, 2022. pp. 661-668 (Frontiers in Artificial Intelligence and Applications).

@inproceedings{b1e831f8aa334f1185424bb007ca05c3,

title = "TRCA: Text Restoration for Chinese ASR with BERT",

abstract = "Text restoration plays a vital role in Chinese automatic speech recognition (ASR), which includes punctuation prediction and error correction. However, there are two inevitable challenges for this task. On the one hand, there are no public dataset and model for Chinese punctuation prediction. On the other hand, current text restoration methods for automatic speech recognition only focus on Chinese error correction instead of combining with Chinese punctuation prediction task. To address these problems, a BERT-based text restoration method called TRCA is proposed for Chinese ASR consisting of a Chinese punctuation prediction model and a Chinese error correction model. Experiments demonstrate that the proposed TRCA method outperforms state-of-the-art methods for both punctuation prediction and error correction tasks, among which the proposed TRCA improves the average accuracy to 98\% in Chinese punctuation prediction.",

keywords = "BERT, Chinese ASR text restoration, Chinese error correction, Chinese punctuation prediction",

author = "Xing Wu and Yuan Zhang and Jianjia Wang and Yike Guo",

note = "Publisher Copyright: {\textcopyright} 2022 The authors and IOS Press. All rights reserved.; 21st International Conference on New Trends in Intelligent Software Methodologies, Tools and Techniques, SoMeT 2022 ; Conference date: 20-09-2022 Through 22-09-2022",

year = "2022",

month = sep,

day = "14",

doi = "10.3233/FAIA220295",

language = "English",

isbn = "9781643683164",

series = "Frontiers in Artificial Intelligence and Applications",

publisher = "IOS Press BV",

pages = "661--668",

editor = "Hamido Fujita and Yutaka Watanobe and Takuya Azumi",

booktitle = "New Trends in Intelligent Software Methodologies, Tools and Techniques - Proceedings of the 21st International Conference on New Trends in Intelligent Software Methodologies, Tools and Techniques, SoMeT 2022",

address = "Netherlands",

}

Wu, X, Zhang, Y, Wang, J & Guo, Y 2022, TRCA: Text Restoration for Chinese ASR with BERT. in H Fujita, Y Watanobe & T Azumi (eds), New Trends in Intelligent Software Methodologies, Tools and Techniques - Proceedings of the 21st International Conference on New Trends in Intelligent Software Methodologies, Tools and Techniques, SoMeT 2022. Frontiers in Artificial Intelligence and Applications, vol. 355, IOS Press BV, pp. 661-668, 21st International Conference on New Trends in Intelligent Software Methodologies, Tools and Techniques, SoMeT 2022, Kitakyushu, Japan, 20/09/22. https://doi.org/10.3233/FAIA220295

TRCA: Text Restoration for Chinese ASR with BERT. / Wu, Xing; Zhang, Yuan; Wang, Jianjia et al.
New Trends in Intelligent Software Methodologies, Tools and Techniques - Proceedings of the 21st International Conference on New Trends in Intelligent Software Methodologies, Tools and Techniques, SoMeT 2022. ed. / Hamido Fujita; Yutaka Watanobe; Takuya Azumi. IOS Press BV, 2022. p. 661-668 (Frontiers in Artificial Intelligence and Applications; Vol. 355).

Research output: Chapter in book/report/conference proceeding › Conference proceeding › peer-review

TY - GEN

T1 - TRCA

T2 - 21st International Conference on New Trends in Intelligent Software Methodologies, Tools and Techniques, SoMeT 2022

AU - Wu, Xing

AU - Zhang, Yuan

AU - Wang, Jianjia

AU - Guo, Yike

PY - 2022/9/14

Y1 - 2022/9/14

N2 - Text restoration plays a vital role in Chinese automatic speech recognition (ASR), which includes punctuation prediction and error correction. However, there are two inevitable challenges for this task. On the one hand, there are no public dataset and model for Chinese punctuation prediction. On the other hand, current text restoration methods for automatic speech recognition only focus on Chinese error correction instead of combining with Chinese punctuation prediction task. To address these problems, a BERT-based text restoration method called TRCA is proposed for Chinese ASR consisting of a Chinese punctuation prediction model and a Chinese error correction model. Experiments demonstrate that the proposed TRCA method outperforms state-of-the-art methods for both punctuation prediction and error correction tasks, among which the proposed TRCA improves the average accuracy to 98% in Chinese punctuation prediction.

AB - Text restoration plays a vital role in Chinese automatic speech recognition (ASR), which includes punctuation prediction and error correction. However, there are two inevitable challenges for this task. On the one hand, there are no public dataset and model for Chinese punctuation prediction. On the other hand, current text restoration methods for automatic speech recognition only focus on Chinese error correction instead of combining with Chinese punctuation prediction task. To address these problems, a BERT-based text restoration method called TRCA is proposed for Chinese ASR consisting of a Chinese punctuation prediction model and a Chinese error correction model. Experiments demonstrate that the proposed TRCA method outperforms state-of-the-art methods for both punctuation prediction and error correction tasks, among which the proposed TRCA improves the average accuracy to 98% in Chinese punctuation prediction.

KW - BERT

KW - Chinese ASR text restoration

KW - Chinese error correction

KW - Chinese punctuation prediction

UR - https://www.scopus.com/pages/publications/85139786485

U2 - 10.3233/FAIA220295

DO - 10.3233/FAIA220295

M3 - Conference proceeding

AN - SCOPUS:85139786485

SN - 9781643683164

T3 - Frontiers in Artificial Intelligence and Applications

SP - 661

EP - 668

BT - New Trends in Intelligent Software Methodologies, Tools and Techniques - Proceedings of the 21st International Conference on New Trends in Intelligent Software Methodologies, Tools and Techniques, SoMeT 2022

A2 - Fujita, Hamido

A2 - Watanobe, Yutaka

A2 - Azumi, Takuya

PB - IOS Press BV

Y2 - 20 September 2022 through 22 September 2022

ER -

Wu X, Zhang Y, Wang J, Guo Y. TRCA: Text Restoration for Chinese ASR with BERT. In Fujita H, Watanobe Y, Azumi T, editors, New Trends in Intelligent Software Methodologies, Tools and Techniques - Proceedings of the 21st International Conference on New Trends in Intelligent Software Methodologies, Tools and Techniques, SoMeT 2022. IOS Press BV. 2022. p. 661-668. (Frontiers in Artificial Intelligence and Applications). doi: 10.3233/FAIA220295

TRCA: Text Restoration for Chinese ASR with BERT

Abstract

Publication series

Conference

UN SDGs

User-Defined Keywords

Access to Document

Other files and links

Fingerprint

Cite this