Speaker recognition based on transformed line spectral frequencies

Bong Jin Lee; Samuel Kim; Hong Goo Kang

Speaker recognition based on transformed line spectral frequencies

Bong Jin Lee, Samuel Kim, Hong Goo Kang

Department of Electrical and Electronic Engineering

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

10 Citations (Scopus)

Abstract

In this paper, line spectral frequencies (LSF) and five types of transformed LSF are studied for robust text-independent speaker identification. Transformations are constructed by considering the physical aspect of vocal tract. They are location of formants/nulls, bandwidth of formants/nulls, bandwidth and location of formants, bandwidth and location of nulls, and interval of adjacent formant and null location. Identification tests using TIMIT database verify that all features are useful for speaker recognition, especially the bandwidth and location of formants show the best performance. Simulation results also show that LSF and some of transformed LSF give better performance than MFCC.

Original language	English
Title of host publication	Proceedings of 2004 International Symposium on Intelligent Signal Processing and Communication Systems, ISPACS 2004
Editors	S.J. Ko
Pages	177-180
Number of pages	4
Publication status	Published - 2004
Event	Proceedings of 2004 International Symposium on Intelligent Signal Processing and Communication Systems, ISPACS 2004 - Seoul, Korea, Republic of Duration: 2004 Nov 18 → 2004 Nov 19

Publication series

Name	Proceedings of 2004 International Symposium on Intelligent Signal Processing and Communication Systems, ISPACS 2004

Other

Other	Proceedings of 2004 International Symposium on Intelligent Signal Processing and Communication Systems, ISPACS 2004
Country/Territory	Korea, Republic of
City	Seoul
Period	04/11/18 → 04/11/19

All Science Journal Classification (ASJC) codes

Engineering(all)

Cite this

Lee, B. J., Kim, S., & Kang, H. G. (2004). Speaker recognition based on transformed line spectral frequencies. In S. J. Ko (Ed.), Proceedings of 2004 International Symposium on Intelligent Signal Processing and Communication Systems, ISPACS 2004 (pp. 177-180). (Proceedings of 2004 International Symposium on Intelligent Signal Processing and Communication Systems, ISPACS 2004).

Lee, Bong Jin ; Kim, Samuel ; Kang, Hong Goo. / Speaker recognition based on transformed line spectral frequencies. Proceedings of 2004 International Symposium on Intelligent Signal Processing and Communication Systems, ISPACS 2004. editor / S.J. Ko. 2004. pp. 177-180 (Proceedings of 2004 International Symposium on Intelligent Signal Processing and Communication Systems, ISPACS 2004).

@inproceedings{4c16a7de03bf4284b9f20829e80294ed,

title = "Speaker recognition based on transformed line spectral frequencies",

abstract = "In this paper, line spectral frequencies (LSF) and five types of transformed LSF are studied for robust text-independent speaker identification. Transformations are constructed by considering the physical aspect of vocal tract. They are location of formants/nulls, bandwidth of formants/nulls, bandwidth and location of formants, bandwidth and location of nulls, and interval of adjacent formant and null location. Identification tests using TIMIT database verify that all features are useful for speaker recognition, especially the bandwidth and location of formants show the best performance. Simulation results also show that LSF and some of transformed LSF give better performance than MFCC.",

author = "Lee, {Bong Jin} and Samuel Kim and Kang, {Hong Goo}",

year = "2004",

language = "English",

isbn = "0780386396",

series = "Proceedings of 2004 International Symposium on Intelligent Signal Processing and Communication Systems, ISPACS 2004",

pages = "177--180",

editor = "S.J. Ko",

booktitle = "Proceedings of 2004 International Symposium on Intelligent Signal Processing and Communication Systems, ISPACS 2004",

note = "Proceedings of 2004 International Symposium on Intelligent Signal Processing and Communication Systems, ISPACS 2004 ; Conference date: 18-11-2004 Through 19-11-2004",

}

Lee, BJ, Kim, S & Kang, HG 2004, Speaker recognition based on transformed line spectral frequencies. in SJ Ko (ed.), Proceedings of 2004 International Symposium on Intelligent Signal Processing and Communication Systems, ISPACS 2004. Proceedings of 2004 International Symposium on Intelligent Signal Processing and Communication Systems, ISPACS 2004, pp. 177-180, Proceedings of 2004 International Symposium on Intelligent Signal Processing and Communication Systems, ISPACS 2004, Seoul, Korea, Republic of, 04/11/18.

Speaker recognition based on transformed line spectral frequencies. / Lee, Bong Jin; Kim, Samuel; Kang, Hong Goo.
Proceedings of 2004 International Symposium on Intelligent Signal Processing and Communication Systems, ISPACS 2004. ed. / S.J. Ko. 2004. p. 177-180 (Proceedings of 2004 International Symposium on Intelligent Signal Processing and Communication Systems, ISPACS 2004).

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

TY - GEN

T1 - Speaker recognition based on transformed line spectral frequencies

AU - Lee, Bong Jin

AU - Kim, Samuel

AU - Kang, Hong Goo

PY - 2004

Y1 - 2004

N2 - In this paper, line spectral frequencies (LSF) and five types of transformed LSF are studied for robust text-independent speaker identification. Transformations are constructed by considering the physical aspect of vocal tract. They are location of formants/nulls, bandwidth of formants/nulls, bandwidth and location of formants, bandwidth and location of nulls, and interval of adjacent formant and null location. Identification tests using TIMIT database verify that all features are useful for speaker recognition, especially the bandwidth and location of formants show the best performance. Simulation results also show that LSF and some of transformed LSF give better performance than MFCC.

AB - In this paper, line spectral frequencies (LSF) and five types of transformed LSF are studied for robust text-independent speaker identification. Transformations are constructed by considering the physical aspect of vocal tract. They are location of formants/nulls, bandwidth of formants/nulls, bandwidth and location of formants, bandwidth and location of nulls, and interval of adjacent formant and null location. Identification tests using TIMIT database verify that all features are useful for speaker recognition, especially the bandwidth and location of formants show the best performance. Simulation results also show that LSF and some of transformed LSF give better performance than MFCC.

UR - http://www.scopus.com/inward/record.url?scp=21544467018&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=21544467018&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:21544467018

SN - 0780386396

T3 - Proceedings of 2004 International Symposium on Intelligent Signal Processing and Communication Systems, ISPACS 2004

SP - 177

EP - 180

BT - Proceedings of 2004 International Symposium on Intelligent Signal Processing and Communication Systems, ISPACS 2004

A2 - Ko, S.J.

T2 - Proceedings of 2004 International Symposium on Intelligent Signal Processing and Communication Systems, ISPACS 2004

Y2 - 18 November 2004 through 19 November 2004

ER -

Speaker recognition based on transformed line spectral frequencies

Abstract

Publication series

Other

All Science Journal Classification (ASJC) codes

Other files and links

Fingerprint

Cite this