Development of small footprint korean large vocabulary speech recognition for commanding a standalone robot

Donghyun Lee; Minkyu Lim; Myoung Wan Koo; Jungyun Seo; Gil Jin Jang; Ji Hwan Kim; Jeong Sik Park

doi:10.1109/BWCCA.2014.112

Development of small footprint korean large vocabulary speech recognition for commanding a standalone robot

Donghyun Lee, Minkyu Lim, Myoung Wan Koo, Jungyun Seo, Gil Jin Jang, Ji Hwan Kim, Jeong Sik Park

School of Electronics Engineering

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

Abstract

The work in this paper concerns a small footprint Acoustic Model (AM) and its use in the implementation of a Large Vocabulary Isolated Speech Recognition (LVISR) system for commanding a robot in the Korean language, which requires about 500KB of memory. Tree-based state clustering was applied to reduce the number of total unique states, while preserving its original performance. A decision tree induction method was developed for the tree-based state clustering. For this method, a binary question set, measurement function and stopping criterion were devised. A phoneme set consisting of 38 phonemes was defined for the implementation of small footprint Korean LVISR. Further reduction in memory requirement was achieved through integer arithmetic operation. The best multiplication factor was determined for this operation. As a result, we successfully developed a small footprint Korean LVISR that requires memory space about 500KB.

Original language	English
Title of host publication	Proceedings - 2014 9th International Conference on Broadband and Wireless Computing, Communication and Applications, BWCCA 2014
Editors	Xiaofeng Chen, Makoto Ikeda, Leonard Barolli, Fatos Xhafa
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	536-540
Number of pages	5
ISBN (Electronic)	9781479941735
DOIs	https://doi.org/10.1109/BWCCA.2014.112
State	Published - 20 Jan 2014
Event	9th IEEE International Conference on Broadband and Wireless Computing, Communication and Applications, BWCCA 2014 - Guangzhou, Guangdong, China Duration: 8 Nov 2014 → 10 Nov 2014

Publication series

Name	Proceedings - 2014 9th International Conference on Broadband and Wireless Computing, Communication and Applications, BWCCA 2014

Conference

Conference	9th IEEE International Conference on Broadband and Wireless Computing, Communication and Applications, BWCCA 2014
Country/Territory	China
City	Guangzhou, Guangdong
Period	8/11/14 → 10/11/14

Keywords

Korean large vocabulary speech recognition
small footprint
standalone robot

Access to Document

10.1109/BWCCA.2014.112

Cite this

Lee, D., Lim, M., Koo, M. W., Seo, J., Jang, G. J., Kim, J. H., & Park, J. S. (2014). Development of small footprint korean large vocabulary speech recognition for commanding a standalone robot. In X. Chen, M. Ikeda, L. Barolli, & F. Xhafa (Eds.), Proceedings - 2014 9th International Conference on Broadband and Wireless Computing, Communication and Applications, BWCCA 2014 (pp. 536-540). Article 7016130 (Proceedings - 2014 9th International Conference on Broadband and Wireless Computing, Communication and Applications, BWCCA 2014). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/BWCCA.2014.112

Lee, Donghyun ; Lim, Minkyu ; Koo, Myoung Wan et al. / Development of small footprint korean large vocabulary speech recognition for commanding a standalone robot. Proceedings - 2014 9th International Conference on Broadband and Wireless Computing, Communication and Applications, BWCCA 2014. editor / Xiaofeng Chen ; Makoto Ikeda ; Leonard Barolli ; Fatos Xhafa. Institute of Electrical and Electronics Engineers Inc., 2014. pp. 536-540 (Proceedings - 2014 9th International Conference on Broadband and Wireless Computing, Communication and Applications, BWCCA 2014).

@inproceedings{9815418a1f3140d0b29f3272efc4096a,

title = "Development of small footprint korean large vocabulary speech recognition for commanding a standalone robot",

abstract = "The work in this paper concerns a small footprint Acoustic Model (AM) and its use in the implementation of a Large Vocabulary Isolated Speech Recognition (LVISR) system for commanding a robot in the Korean language, which requires about 500KB of memory. Tree-based state clustering was applied to reduce the number of total unique states, while preserving its original performance. A decision tree induction method was developed for the tree-based state clustering. For this method, a binary question set, measurement function and stopping criterion were devised. A phoneme set consisting of 38 phonemes was defined for the implementation of small footprint Korean LVISR. Further reduction in memory requirement was achieved through integer arithmetic operation. The best multiplication factor was determined for this operation. As a result, we successfully developed a small footprint Korean LVISR that requires memory space about 500KB.",

keywords = "Korean large vocabulary speech recognition, small footprint, standalone robot",

author = "Donghyun Lee and Minkyu Lim and Koo, {Myoung Wan} and Jungyun Seo and Jang, {Gil Jin} and Kim, {Ji Hwan} and Park, {Jeong Sik}",

note = "Publisher Copyright: {\textcopyright} 2014 IEEE.; 9th IEEE International Conference on Broadband and Wireless Computing, Communication and Applications, BWCCA 2014 ; Conference date: 08-11-2014 Through 10-11-2014",

year = "2014",

month = jan,

day = "20",

doi = "10.1109/BWCCA.2014.112",

language = "English",

series = "Proceedings - 2014 9th International Conference on Broadband and Wireless Computing, Communication and Applications, BWCCA 2014",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "536--540",

editor = "Xiaofeng Chen and Makoto Ikeda and Leonard Barolli and Fatos Xhafa",

booktitle = "Proceedings - 2014 9th International Conference on Broadband and Wireless Computing, Communication and Applications, BWCCA 2014",

address = "United States",

}

Lee, D, Lim, M, Koo, MW, Seo, J, Jang, GJ, Kim, JH & Park, JS 2014, Development of small footprint korean large vocabulary speech recognition for commanding a standalone robot. in X Chen, M Ikeda, L Barolli & F Xhafa (eds), Proceedings - 2014 9th International Conference on Broadband and Wireless Computing, Communication and Applications, BWCCA 2014., 7016130, Proceedings - 2014 9th International Conference on Broadband and Wireless Computing, Communication and Applications, BWCCA 2014, Institute of Electrical and Electronics Engineers Inc., pp. 536-540, 9th IEEE International Conference on Broadband and Wireless Computing, Communication and Applications, BWCCA 2014, Guangzhou, Guangdong, China, 8/11/14. https://doi.org/10.1109/BWCCA.2014.112

Development of small footprint korean large vocabulary speech recognition for commanding a standalone robot. / Lee, Donghyun; Lim, Minkyu; Koo, Myoung Wan et al.
Proceedings - 2014 9th International Conference on Broadband and Wireless Computing, Communication and Applications, BWCCA 2014. ed. / Xiaofeng Chen; Makoto Ikeda; Leonard Barolli; Fatos Xhafa. Institute of Electrical and Electronics Engineers Inc., 2014. p. 536-540 7016130 (Proceedings - 2014 9th International Conference on Broadband and Wireless Computing, Communication and Applications, BWCCA 2014).

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

TY - GEN

T1 - Development of small footprint korean large vocabulary speech recognition for commanding a standalone robot

AU - Lee, Donghyun

AU - Lim, Minkyu

AU - Koo, Myoung Wan

AU - Seo, Jungyun

AU - Jang, Gil Jin

AU - Kim, Ji Hwan

AU - Park, Jeong Sik

PY - 2014/1/20

Y1 - 2014/1/20

N2 - The work in this paper concerns a small footprint Acoustic Model (AM) and its use in the implementation of a Large Vocabulary Isolated Speech Recognition (LVISR) system for commanding a robot in the Korean language, which requires about 500KB of memory. Tree-based state clustering was applied to reduce the number of total unique states, while preserving its original performance. A decision tree induction method was developed for the tree-based state clustering. For this method, a binary question set, measurement function and stopping criterion were devised. A phoneme set consisting of 38 phonemes was defined for the implementation of small footprint Korean LVISR. Further reduction in memory requirement was achieved through integer arithmetic operation. The best multiplication factor was determined for this operation. As a result, we successfully developed a small footprint Korean LVISR that requires memory space about 500KB.

AB - The work in this paper concerns a small footprint Acoustic Model (AM) and its use in the implementation of a Large Vocabulary Isolated Speech Recognition (LVISR) system for commanding a robot in the Korean language, which requires about 500KB of memory. Tree-based state clustering was applied to reduce the number of total unique states, while preserving its original performance. A decision tree induction method was developed for the tree-based state clustering. For this method, a binary question set, measurement function and stopping criterion were devised. A phoneme set consisting of 38 phonemes was defined for the implementation of small footprint Korean LVISR. Further reduction in memory requirement was achieved through integer arithmetic operation. The best multiplication factor was determined for this operation. As a result, we successfully developed a small footprint Korean LVISR that requires memory space about 500KB.

KW - Korean large vocabulary speech recognition

KW - small footprint

KW - standalone robot

UR - http://www.scopus.com/inward/record.url?scp=84946687876&partnerID=8YFLogxK

U2 - 10.1109/BWCCA.2014.112

DO - 10.1109/BWCCA.2014.112

M3 - Conference contribution

AN - SCOPUS:84946687876

T3 - Proceedings - 2014 9th International Conference on Broadband and Wireless Computing, Communication and Applications, BWCCA 2014

SP - 536

EP - 540

BT - Proceedings - 2014 9th International Conference on Broadband and Wireless Computing, Communication and Applications, BWCCA 2014

A2 - Chen, Xiaofeng

A2 - Ikeda, Makoto

A2 - Barolli, Leonard

A2 - Xhafa, Fatos

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 9th IEEE International Conference on Broadband and Wireless Computing, Communication and Applications, BWCCA 2014

Y2 - 8 November 2014 through 10 November 2014

ER -

Lee D, Lim M, Koo MW, Seo J, Jang GJ, Kim JH et al. Development of small footprint korean large vocabulary speech recognition for commanding a standalone robot. In Chen X, Ikeda M, Barolli L, Xhafa F, editors, Proceedings - 2014 9th International Conference on Broadband and Wireless Computing, Communication and Applications, BWCCA 2014. Institute of Electrical and Electronics Engineers Inc. 2014. p. 536-540. 7016130. (Proceedings - 2014 9th International Conference on Broadband and Wireless Computing, Communication and Applications, BWCCA 2014). doi: 10.1109/BWCCA.2014.112

Development of small footprint korean large vocabulary speech recognition for commanding a standalone robot

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this