Exploring kernel discriminant analysis for speaker verification with limited test data

•A novel framework for channel/session compensation in i-vector speaker modeling.•Explore non-linearity in channel/session information at i-vector framework.•Effectiveness of kernel discriminant analysis (KDA) with higher dimension.•Significance of KDA for speaker verification with limited test data...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Pattern recognition letters 2017-10, Vol.98, p.26-31
Hauptverfasser: Das, Rohan Kumar, Manam, Akhil Babu, Prasanna, S.R. Mahadeva
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page 31
container_issue
container_start_page 26
container_title Pattern recognition letters
container_volume 98
creator Das, Rohan Kumar
Manam, Akhil Babu
Prasanna, S.R. Mahadeva
description •A novel framework for channel/session compensation in i-vector speaker modeling.•Explore non-linearity in channel/session information at i-vector framework.•Effectiveness of kernel discriminant analysis (KDA) with higher dimension.•Significance of KDA for speaker verification with limited test data. Speaker verification (SV) with limited test data condition is desirable for practical application oriented systems. The i-vector based speaker modeling has shown its significance for SV tasks, but its performance degrades as the utterance becomes shorter. The i-vectors apart from being compact and dominant speaker representations, bear channel and session information, which has to be compensated for robust speaker modeling. The conventional techniques for channel/session compensation include linear discriminant analysis (LDA) followed by within class covariance normalization (WCCN) and Gaussian probabilistic linear discriminant analysis (GPLDA) that eliminate the channel/session variation across the i-vectors by assuming these are linearly separable. In this work, a novel method for channel/session compensation is proposed using kernel discriminant analysis (KDA) that projects the i-vectors into a higher dimensional space and performs discriminant analysis to remove the unwanted information for speaker modeling. The SV studies are performed on standard NIST speaker recognition evaluation (SRE) 2003 and 2008 databases that convey the significance of the proposed compensation over the conventional methods, which is greater on using short test utterances. The achieved improvements are hypothesized due to the non-linearities of channel/session information in the i-vector domain.
doi_str_mv 10.1016/j.patrec.2017.08.004
format Article
fullrecord <record><control><sourceid>proquest_cross</sourceid><recordid>TN_cdi_proquest_journals_2089732684</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><els_id>S0167865517302611</els_id><sourcerecordid>2089732684</sourcerecordid><originalsourceid>FETCH-LOGICAL-c334t-ac85503205877235f92472bb66942e1eeeb3921e254853678dfab09d12ae7f793</originalsourceid><addsrcrecordid>eNp9kMtOwzAQRS0EEqXwBywssU7wK7GzQUJVeUiV2ABby3Em4BCSYLuF_j2uwprVbM69M3MQuqQkp4SW110-mejB5oxQmROVEyKO0IIqyTLJhThGi4TJTJVFcYrOQugIISWv1AK9rn-mfvRueMMf4AfoceOC9e7TDWaI2Aym3wcXcDt6HCYwCcI78K511kQ3DvjbxXfcJz5CgyOEiBsTzTk6aU0f4OJvLtHL3fp59ZBtnu4fV7ebzHIuYmasKgrCGSmUlIwXbcWEZHVdlpVgQAGg5hWjwAqhCl5K1bSmJlVDmQHZyoov0dXcO_nxa5u2627c-nR00IyoSnJWKpEoMVPWjyF4aPWUPjR-rynRB4O607NBfTCoidLJYIrdzDFIH-wceB2sg8FC4xIadTO6_wt-AfLjfAU</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2089732684</pqid></control><display><type>article</type><title>Exploring kernel discriminant analysis for speaker verification with limited test data</title><source>Elsevier ScienceDirect Journals Complete</source><creator>Das, Rohan Kumar ; Manam, Akhil Babu ; Prasanna, S.R. Mahadeva</creator><creatorcontrib>Das, Rohan Kumar ; Manam, Akhil Babu ; Prasanna, S.R. Mahadeva</creatorcontrib><description>•A novel framework for channel/session compensation in i-vector speaker modeling.•Explore non-linearity in channel/session information at i-vector framework.•Effectiveness of kernel discriminant analysis (KDA) with higher dimension.•Significance of KDA for speaker verification with limited test data. Speaker verification (SV) with limited test data condition is desirable for practical application oriented systems. The i-vector based speaker modeling has shown its significance for SV tasks, but its performance degrades as the utterance becomes shorter. The i-vectors apart from being compact and dominant speaker representations, bear channel and session information, which has to be compensated for robust speaker modeling. The conventional techniques for channel/session compensation include linear discriminant analysis (LDA) followed by within class covariance normalization (WCCN) and Gaussian probabilistic linear discriminant analysis (GPLDA) that eliminate the channel/session variation across the i-vectors by assuming these are linearly separable. In this work, a novel method for channel/session compensation is proposed using kernel discriminant analysis (KDA) that projects the i-vectors into a higher dimensional space and performs discriminant analysis to remove the unwanted information for speaker modeling. The SV studies are performed on standard NIST speaker recognition evaluation (SRE) 2003 and 2008 databases that convey the significance of the proposed compensation over the conventional methods, which is greater on using short test utterances. The achieved improvements are hypothesized due to the non-linearities of channel/session information in the i-vector domain.</description><identifier>ISSN: 0167-8655</identifier><identifier>EISSN: 1872-7344</identifier><identifier>DOI: 10.1016/j.patrec.2017.08.004</identifier><language>eng</language><publisher>Amsterdam: Elsevier B.V</publisher><subject>Array processors ; Compensation ; Covariance ; Data processing ; Discriminant analysis ; Gaussian distribution ; I-vector ; Kernel discriminant analysis (KDA) ; KSVD ; Linear programming ; Modelling ; Pattern recognition ; Performance degradation ; Short utterance ; Speaker verification (SV) ; Speech recognition ; Voice recognition</subject><ispartof>Pattern recognition letters, 2017-10, Vol.98, p.26-31</ispartof><rights>2017 Elsevier B.V.</rights><rights>Copyright Elsevier Science Ltd. Oct 15, 2017</rights><lds50>peer_reviewed</lds50><woscitedreferencessubscribed>false</woscitedreferencessubscribed><citedby>FETCH-LOGICAL-c334t-ac85503205877235f92472bb66942e1eeeb3921e254853678dfab09d12ae7f793</citedby><cites>FETCH-LOGICAL-c334t-ac85503205877235f92472bb66942e1eeeb3921e254853678dfab09d12ae7f793</cites></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://dx.doi.org/10.1016/j.patrec.2017.08.004$$EHTML$$P50$$Gelsevier$$H</linktohtml><link.rule.ids>314,780,784,3550,27924,27925,45995</link.rule.ids></links><search><creatorcontrib>Das, Rohan Kumar</creatorcontrib><creatorcontrib>Manam, Akhil Babu</creatorcontrib><creatorcontrib>Prasanna, S.R. Mahadeva</creatorcontrib><title>Exploring kernel discriminant analysis for speaker verification with limited test data</title><title>Pattern recognition letters</title><description>•A novel framework for channel/session compensation in i-vector speaker modeling.•Explore non-linearity in channel/session information at i-vector framework.•Effectiveness of kernel discriminant analysis (KDA) with higher dimension.•Significance of KDA for speaker verification with limited test data. Speaker verification (SV) with limited test data condition is desirable for practical application oriented systems. The i-vector based speaker modeling has shown its significance for SV tasks, but its performance degrades as the utterance becomes shorter. The i-vectors apart from being compact and dominant speaker representations, bear channel and session information, which has to be compensated for robust speaker modeling. The conventional techniques for channel/session compensation include linear discriminant analysis (LDA) followed by within class covariance normalization (WCCN) and Gaussian probabilistic linear discriminant analysis (GPLDA) that eliminate the channel/session variation across the i-vectors by assuming these are linearly separable. In this work, a novel method for channel/session compensation is proposed using kernel discriminant analysis (KDA) that projects the i-vectors into a higher dimensional space and performs discriminant analysis to remove the unwanted information for speaker modeling. The SV studies are performed on standard NIST speaker recognition evaluation (SRE) 2003 and 2008 databases that convey the significance of the proposed compensation over the conventional methods, which is greater on using short test utterances. The achieved improvements are hypothesized due to the non-linearities of channel/session information in the i-vector domain.</description><subject>Array processors</subject><subject>Compensation</subject><subject>Covariance</subject><subject>Data processing</subject><subject>Discriminant analysis</subject><subject>Gaussian distribution</subject><subject>I-vector</subject><subject>Kernel discriminant analysis (KDA)</subject><subject>KSVD</subject><subject>Linear programming</subject><subject>Modelling</subject><subject>Pattern recognition</subject><subject>Performance degradation</subject><subject>Short utterance</subject><subject>Speaker verification (SV)</subject><subject>Speech recognition</subject><subject>Voice recognition</subject><issn>0167-8655</issn><issn>1872-7344</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2017</creationdate><recordtype>article</recordtype><recordid>eNp9kMtOwzAQRS0EEqXwBywssU7wK7GzQUJVeUiV2ABby3Em4BCSYLuF_j2uwprVbM69M3MQuqQkp4SW110-mejB5oxQmROVEyKO0IIqyTLJhThGi4TJTJVFcYrOQugIISWv1AK9rn-mfvRueMMf4AfoceOC9e7TDWaI2Aym3wcXcDt6HCYwCcI78K511kQ3DvjbxXfcJz5CgyOEiBsTzTk6aU0f4OJvLtHL3fp59ZBtnu4fV7ebzHIuYmasKgrCGSmUlIwXbcWEZHVdlpVgQAGg5hWjwAqhCl5K1bSmJlVDmQHZyoov0dXcO_nxa5u2627c-nR00IyoSnJWKpEoMVPWjyF4aPWUPjR-rynRB4O607NBfTCoidLJYIrdzDFIH-wceB2sg8FC4xIadTO6_wt-AfLjfAU</recordid><startdate>20171015</startdate><enddate>20171015</enddate><creator>Das, Rohan Kumar</creator><creator>Manam, Akhil Babu</creator><creator>Prasanna, S.R. Mahadeva</creator><general>Elsevier B.V</general><general>Elsevier Science Ltd</general><scope>AAYXX</scope><scope>CITATION</scope><scope>7SC</scope><scope>7TK</scope><scope>8FD</scope><scope>JQ2</scope><scope>L7M</scope><scope>L~C</scope><scope>L~D</scope></search><sort><creationdate>20171015</creationdate><title>Exploring kernel discriminant analysis for speaker verification with limited test data</title><author>Das, Rohan Kumar ; Manam, Akhil Babu ; Prasanna, S.R. Mahadeva</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c334t-ac85503205877235f92472bb66942e1eeeb3921e254853678dfab09d12ae7f793</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2017</creationdate><topic>Array processors</topic><topic>Compensation</topic><topic>Covariance</topic><topic>Data processing</topic><topic>Discriminant analysis</topic><topic>Gaussian distribution</topic><topic>I-vector</topic><topic>Kernel discriminant analysis (KDA)</topic><topic>KSVD</topic><topic>Linear programming</topic><topic>Modelling</topic><topic>Pattern recognition</topic><topic>Performance degradation</topic><topic>Short utterance</topic><topic>Speaker verification (SV)</topic><topic>Speech recognition</topic><topic>Voice recognition</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Das, Rohan Kumar</creatorcontrib><creatorcontrib>Manam, Akhil Babu</creatorcontrib><creatorcontrib>Prasanna, S.R. Mahadeva</creatorcontrib><collection>CrossRef</collection><collection>Computer and Information Systems Abstracts</collection><collection>Neurosciences Abstracts</collection><collection>Technology Research Database</collection><collection>ProQuest Computer Science Collection</collection><collection>Advanced Technologies Database with Aerospace</collection><collection>Computer and Information Systems Abstracts – Academic</collection><collection>Computer and Information Systems Abstracts Professional</collection><jtitle>Pattern recognition letters</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Das, Rohan Kumar</au><au>Manam, Akhil Babu</au><au>Prasanna, S.R. Mahadeva</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Exploring kernel discriminant analysis for speaker verification with limited test data</atitle><jtitle>Pattern recognition letters</jtitle><date>2017-10-15</date><risdate>2017</risdate><volume>98</volume><spage>26</spage><epage>31</epage><pages>26-31</pages><issn>0167-8655</issn><eissn>1872-7344</eissn><abstract>•A novel framework for channel/session compensation in i-vector speaker modeling.•Explore non-linearity in channel/session information at i-vector framework.•Effectiveness of kernel discriminant analysis (KDA) with higher dimension.•Significance of KDA for speaker verification with limited test data. Speaker verification (SV) with limited test data condition is desirable for practical application oriented systems. The i-vector based speaker modeling has shown its significance for SV tasks, but its performance degrades as the utterance becomes shorter. The i-vectors apart from being compact and dominant speaker representations, bear channel and session information, which has to be compensated for robust speaker modeling. The conventional techniques for channel/session compensation include linear discriminant analysis (LDA) followed by within class covariance normalization (WCCN) and Gaussian probabilistic linear discriminant analysis (GPLDA) that eliminate the channel/session variation across the i-vectors by assuming these are linearly separable. In this work, a novel method for channel/session compensation is proposed using kernel discriminant analysis (KDA) that projects the i-vectors into a higher dimensional space and performs discriminant analysis to remove the unwanted information for speaker modeling. The SV studies are performed on standard NIST speaker recognition evaluation (SRE) 2003 and 2008 databases that convey the significance of the proposed compensation over the conventional methods, which is greater on using short test utterances. The achieved improvements are hypothesized due to the non-linearities of channel/session information in the i-vector domain.</abstract><cop>Amsterdam</cop><pub>Elsevier B.V</pub><doi>10.1016/j.patrec.2017.08.004</doi><tpages>6</tpages></addata></record>
fulltext fulltext
identifier ISSN: 0167-8655
ispartof Pattern recognition letters, 2017-10, Vol.98, p.26-31
issn 0167-8655
1872-7344
language eng
recordid cdi_proquest_journals_2089732684
source Elsevier ScienceDirect Journals Complete
subjects Array processors
Compensation
Covariance
Data processing
Discriminant analysis
Gaussian distribution
I-vector
Kernel discriminant analysis (KDA)
KSVD
Linear programming
Modelling
Pattern recognition
Performance degradation
Short utterance
Speaker verification (SV)
Speech recognition
Voice recognition
title Exploring kernel discriminant analysis for speaker verification with limited test data
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-25T10%3A35%3A42IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest_cross&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Exploring%20kernel%20discriminant%20analysis%20for%20speaker%20verification%20with%20limited%20test%20data&rft.jtitle=Pattern%20recognition%20letters&rft.au=Das,%20Rohan%20Kumar&rft.date=2017-10-15&rft.volume=98&rft.spage=26&rft.epage=31&rft.pages=26-31&rft.issn=0167-8655&rft.eissn=1872-7344&rft_id=info:doi/10.1016/j.patrec.2017.08.004&rft_dat=%3Cproquest_cross%3E2089732684%3C/proquest_cross%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=2089732684&rft_id=info:pmid/&rft_els_id=S0167865517302611&rfr_iscdi=true