Speaker recognition in unknown mismatched conditions using augmented PCA

Ha Jin Yu

doi:10.1007/11569596_69

Speaker recognition in unknown mismatched conditions using augmented PCA

Ha Jin Yu

Department of Computer Science and Engineering

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

Abstract

Our goal was to build a text-independent speaker recognition system that could be used under any conditions without any additional adaptation process. Unknown mismatched microphones and noise conditions can severely degrade the performance of speaker recognition systems. This paper shows that principal component analysis (PCA) can increase performance under these conditions without reducing dimension. We also propose a PCA process that augments class discriminative information sent to original feature vectors before PCA transformation and selects the best direction between each pair of highly confusable speakers. In tests, the proposed method reduced errors in recognition by 32%.

Original language	English
Title of host publication	Computer and Information Sciences - ISCIS 2005 - 20th International Symposium, Proceedings
Pages	668-676
Number of pages	9
DOIs	https://doi.org/10.1007/11569596_69
State	Published - 2005
Event	20th International Symposium on Computer and Information Sciences, ISCIS 2005 - Istanbul, Turkey Duration: 26 Oct 2005 → 28 Oct 2005

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	3733 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	20th International Symposium on Computer and Information Sciences, ISCIS 2005
Country/Territory	Turkey
City	Istanbul
Period	26/10/05 → 28/10/05

Access to Document

10.1007/11569596_69

Cite this

Yu, H. J. (2005). Speaker recognition in unknown mismatched conditions using augmented PCA. In Computer and Information Sciences - ISCIS 2005 - 20th International Symposium, Proceedings (pp. 668-676). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 3733 LNCS). https://doi.org/10.1007/11569596_69

@inproceedings{7f1344ea8fcf4ee28ce9921571641168,

title = "Speaker recognition in unknown mismatched conditions using augmented PCA",

abstract = "Our goal was to build a text-independent speaker recognition system that could be used under any conditions without any additional adaptation process. Unknown mismatched microphones and noise conditions can severely degrade the performance of speaker recognition systems. This paper shows that principal component analysis (PCA) can increase performance under these conditions without reducing dimension. We also propose a PCA process that augments class discriminative information sent to original feature vectors before PCA transformation and selects the best direction between each pair of highly confusable speakers. In tests, the proposed method reduced errors in recognition by 32%.",

author = "Yu, {Ha Jin}",

year = "2005",

doi = "10.1007/11569596_69",

language = "English",

isbn = "3540294147",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

pages = "668--676",

booktitle = "Computer and Information Sciences - ISCIS 2005 - 20th International Symposium, Proceedings",

note = "20th International Symposium on Computer and Information Sciences, ISCIS 2005 ; Conference date: 26-10-2005 Through 28-10-2005",

}

Yu, HJ 2005, Speaker recognition in unknown mismatched conditions using augmented PCA. in Computer and Information Sciences - ISCIS 2005 - 20th International Symposium, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 3733 LNCS, pp. 668-676, 20th International Symposium on Computer and Information Sciences, ISCIS 2005, Istanbul, Turkey, 26/10/05. https://doi.org/10.1007/11569596_69

Speaker recognition in unknown mismatched conditions using augmented PCA. / Yu, Ha Jin.
Computer and Information Sciences - ISCIS 2005 - 20th International Symposium, Proceedings. 2005. p. 668-676 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 3733 LNCS).

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

TY - GEN

T1 - Speaker recognition in unknown mismatched conditions using augmented PCA

AU - Yu, Ha Jin

PY - 2005

Y1 - 2005

N2 - Our goal was to build a text-independent speaker recognition system that could be used under any conditions without any additional adaptation process. Unknown mismatched microphones and noise conditions can severely degrade the performance of speaker recognition systems. This paper shows that principal component analysis (PCA) can increase performance under these conditions without reducing dimension. We also propose a PCA process that augments class discriminative information sent to original feature vectors before PCA transformation and selects the best direction between each pair of highly confusable speakers. In tests, the proposed method reduced errors in recognition by 32%.

AB - Our goal was to build a text-independent speaker recognition system that could be used under any conditions without any additional adaptation process. Unknown mismatched microphones and noise conditions can severely degrade the performance of speaker recognition systems. This paper shows that principal component analysis (PCA) can increase performance under these conditions without reducing dimension. We also propose a PCA process that augments class discriminative information sent to original feature vectors before PCA transformation and selects the best direction between each pair of highly confusable speakers. In tests, the proposed method reduced errors in recognition by 32%.

UR - http://www.scopus.com/inward/record.url?scp=33646502804&partnerID=8YFLogxK

U2 - 10.1007/11569596_69

DO - 10.1007/11569596_69

M3 - Conference contribution

AN - SCOPUS:33646502804

SN - 3540294147

SN - 9783540294146

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 668

EP - 676

BT - Computer and Information Sciences - ISCIS 2005 - 20th International Symposium, Proceedings

T2 - 20th International Symposium on Computer and Information Sciences, ISCIS 2005

Y2 - 26 October 2005 through 28 October 2005

ER -

Speaker recognition in unknown mismatched conditions using augmented PCA

Abstract

Publication series

Conference

Access to Document

Other files and links

Fingerprint

Cite this