Toward Fail-Safe Speaker Recognition: Trial-Based Calibration with a Reject Option

Autores
Ferrer, Luciana; Nandwana, Mahesh Kumar; McLaren, Mitchell; Castan, Diego; Lawson, Aaron
Año de publicación
2019
Idioma
inglés
Tipo de recurso
artículo
Estado
versión publicada
Descripción
The output scores of most of the speaker recognition systems are not directly interpretable as stand-alone values. For this reason, a calibration step is usually performed on the scores to convert them into proper likelihood ratios, which have a clear probabilistic interpretation. The standard calibration approach transforms the system scores using a linear function trained using data selected to closely match the evaluation conditions. This selection, though, is not feasible when the evaluation conditions are unknown. In previous work, we proposed a calibration approach for this scenario called trial-based calibration (TBC). TBC trains a separate calibration model for each test trial using data that is dynamically selected from a candidate training set to match the conditions of the trial. In this work, we extend the TBC method, proposing: 1) a new similarity metric for selecting training data that result in significant gains over the one proposed in the original work; 2) a new option that enables the system to reject a trial when not enough matched data are available for training the calibration model; and 3) the use of regularization to improve the robustness of the calibration models trained for each trial. We test the proposed algorithms on a development set composed of several conditions and on the Federal Bureau of Investigation multi-condition speaker recognition dataset, and we demonstrate that the proposed approach reduces calibration loss to values close to 0 for most of the conditions when matched calibration data are available for selection, and that it can reject most of the trials for which relevant calibration data are unavailable.
Fil: Ferrer, Luciana. Consejo Nacional de Investigaciones Científicas y Técnicas. Oficina de Coordinación Administrativa Ciudad Universitaria. Instituto de Investigación en Ciencias de la Computación. Universidad de Buenos Aires. Facultad de Ciencias Exactas y Naturales. Instituto de Investigación en Ciencias de la Computación; Argentina
Fil: Nandwana, Mahesh Kumar. No especifíca;
Fil: McLaren, Mitchell. No especifíca;
Fil: Castan, Diego. No especifíca;
Fil: Lawson, Aaron. No especifíca;
Materia
FORENSIC VOICE COMPARISON
SPEAKER RECOGNITION
TRIAL-BASED CALIBRATION
Nivel de accesibilidad
acceso abierto
Condiciones de uso
https://creativecommons.org/licenses/by-nc-sa/2.5/ar/
Repositorio
CONICET Digital (CONICET)
Institución
Consejo Nacional de Investigaciones Científicas y Técnicas
OAI Identificador
oai:ri.conicet.gov.ar:11336/123318

id CONICETDig_70a22c3fffd6efdd214a14209300f2b8
oai_identifier_str oai:ri.conicet.gov.ar:11336/123318
network_acronym_str CONICETDig
repository_id_str 3498
network_name_str CONICET Digital (CONICET)
spelling Toward Fail-Safe Speaker Recognition: Trial-Based Calibration with a Reject OptionFerrer, LucianaNandwana, Mahesh KumarMcLaren, MitchellCastan, DiegoLawson, AaronFORENSIC VOICE COMPARISONSPEAKER RECOGNITIONTRIAL-BASED CALIBRATIONhttps://purl.org/becyt/ford/1.2https://purl.org/becyt/ford/1The output scores of most of the speaker recognition systems are not directly interpretable as stand-alone values. For this reason, a calibration step is usually performed on the scores to convert them into proper likelihood ratios, which have a clear probabilistic interpretation. The standard calibration approach transforms the system scores using a linear function trained using data selected to closely match the evaluation conditions. This selection, though, is not feasible when the evaluation conditions are unknown. In previous work, we proposed a calibration approach for this scenario called trial-based calibration (TBC). TBC trains a separate calibration model for each test trial using data that is dynamically selected from a candidate training set to match the conditions of the trial. In this work, we extend the TBC method, proposing: 1) a new similarity metric for selecting training data that result in significant gains over the one proposed in the original work; 2) a new option that enables the system to reject a trial when not enough matched data are available for training the calibration model; and 3) the use of regularization to improve the robustness of the calibration models trained for each trial. We test the proposed algorithms on a development set composed of several conditions and on the Federal Bureau of Investigation multi-condition speaker recognition dataset, and we demonstrate that the proposed approach reduces calibration loss to values close to 0 for most of the conditions when matched calibration data are available for selection, and that it can reject most of the trials for which relevant calibration data are unavailable.Fil: Ferrer, Luciana. Consejo Nacional de Investigaciones Científicas y Técnicas. Oficina de Coordinación Administrativa Ciudad Universitaria. Instituto de Investigación en Ciencias de la Computación. Universidad de Buenos Aires. Facultad de Ciencias Exactas y Naturales. Instituto de Investigación en Ciencias de la Computación; ArgentinaFil: Nandwana, Mahesh Kumar. No especifíca;Fil: McLaren, Mitchell. No especifíca;Fil: Castan, Diego. No especifíca;Fil: Lawson, Aaron. No especifíca;Institute of Electrical and Electronics Engineers2019-01info:eu-repo/semantics/articleinfo:eu-repo/semantics/publishedVersionhttp://purl.org/coar/resource_type/c_6501info:ar-repo/semantics/articuloapplication/pdfapplication/pdfhttp://hdl.handle.net/11336/123318Ferrer, Luciana; Nandwana, Mahesh Kumar; McLaren, Mitchell; Castan, Diego; Lawson, Aaron; Toward Fail-Safe Speaker Recognition: Trial-Based Calibration with a Reject Option; Institute of Electrical and Electronics Engineers; IEEE/ACM Transactions on Audio Speech and Language Processing; 27; 1; 1-2019; 140-1532329-9290CONICET DigitalCONICETenginfo:eu-repo/semantics/altIdentifier/url/https://ieeexplore.ieee.org/document/8490592info:eu-repo/semantics/altIdentifier/doi/10.1109/TASLP.2018.2875794info:eu-repo/semantics/openAccesshttps://creativecommons.org/licenses/by-nc-sa/2.5/ar/reponame:CONICET Digital (CONICET)instname:Consejo Nacional de Investigaciones Científicas y Técnicas2025-09-29T10:36:56Zoai:ri.conicet.gov.ar:11336/123318instacron:CONICETInstitucionalhttp://ri.conicet.gov.ar/Organismo científico-tecnológicoNo correspondehttp://ri.conicet.gov.ar/oai/requestdasensio@conicet.gov.ar; lcarlino@conicet.gov.arArgentinaNo correspondeNo correspondeNo correspondeopendoar:34982025-09-29 10:36:56.339CONICET Digital (CONICET) - Consejo Nacional de Investigaciones Científicas y Técnicasfalse
dc.title.none.fl_str_mv Toward Fail-Safe Speaker Recognition: Trial-Based Calibration with a Reject Option
title Toward Fail-Safe Speaker Recognition: Trial-Based Calibration with a Reject Option
spellingShingle Toward Fail-Safe Speaker Recognition: Trial-Based Calibration with a Reject Option
Ferrer, Luciana
FORENSIC VOICE COMPARISON
SPEAKER RECOGNITION
TRIAL-BASED CALIBRATION
title_short Toward Fail-Safe Speaker Recognition: Trial-Based Calibration with a Reject Option
title_full Toward Fail-Safe Speaker Recognition: Trial-Based Calibration with a Reject Option
title_fullStr Toward Fail-Safe Speaker Recognition: Trial-Based Calibration with a Reject Option
title_full_unstemmed Toward Fail-Safe Speaker Recognition: Trial-Based Calibration with a Reject Option
title_sort Toward Fail-Safe Speaker Recognition: Trial-Based Calibration with a Reject Option
dc.creator.none.fl_str_mv Ferrer, Luciana
Nandwana, Mahesh Kumar
McLaren, Mitchell
Castan, Diego
Lawson, Aaron
author Ferrer, Luciana
author_facet Ferrer, Luciana
Nandwana, Mahesh Kumar
McLaren, Mitchell
Castan, Diego
Lawson, Aaron
author_role author
author2 Nandwana, Mahesh Kumar
McLaren, Mitchell
Castan, Diego
Lawson, Aaron
author2_role author
author
author
author
dc.subject.none.fl_str_mv FORENSIC VOICE COMPARISON
SPEAKER RECOGNITION
TRIAL-BASED CALIBRATION
topic FORENSIC VOICE COMPARISON
SPEAKER RECOGNITION
TRIAL-BASED CALIBRATION
purl_subject.fl_str_mv https://purl.org/becyt/ford/1.2
https://purl.org/becyt/ford/1
dc.description.none.fl_txt_mv The output scores of most of the speaker recognition systems are not directly interpretable as stand-alone values. For this reason, a calibration step is usually performed on the scores to convert them into proper likelihood ratios, which have a clear probabilistic interpretation. The standard calibration approach transforms the system scores using a linear function trained using data selected to closely match the evaluation conditions. This selection, though, is not feasible when the evaluation conditions are unknown. In previous work, we proposed a calibration approach for this scenario called trial-based calibration (TBC). TBC trains a separate calibration model for each test trial using data that is dynamically selected from a candidate training set to match the conditions of the trial. In this work, we extend the TBC method, proposing: 1) a new similarity metric for selecting training data that result in significant gains over the one proposed in the original work; 2) a new option that enables the system to reject a trial when not enough matched data are available for training the calibration model; and 3) the use of regularization to improve the robustness of the calibration models trained for each trial. We test the proposed algorithms on a development set composed of several conditions and on the Federal Bureau of Investigation multi-condition speaker recognition dataset, and we demonstrate that the proposed approach reduces calibration loss to values close to 0 for most of the conditions when matched calibration data are available for selection, and that it can reject most of the trials for which relevant calibration data are unavailable.
Fil: Ferrer, Luciana. Consejo Nacional de Investigaciones Científicas y Técnicas. Oficina de Coordinación Administrativa Ciudad Universitaria. Instituto de Investigación en Ciencias de la Computación. Universidad de Buenos Aires. Facultad de Ciencias Exactas y Naturales. Instituto de Investigación en Ciencias de la Computación; Argentina
Fil: Nandwana, Mahesh Kumar. No especifíca;
Fil: McLaren, Mitchell. No especifíca;
Fil: Castan, Diego. No especifíca;
Fil: Lawson, Aaron. No especifíca;
description The output scores of most of the speaker recognition systems are not directly interpretable as stand-alone values. For this reason, a calibration step is usually performed on the scores to convert them into proper likelihood ratios, which have a clear probabilistic interpretation. The standard calibration approach transforms the system scores using a linear function trained using data selected to closely match the evaluation conditions. This selection, though, is not feasible when the evaluation conditions are unknown. In previous work, we proposed a calibration approach for this scenario called trial-based calibration (TBC). TBC trains a separate calibration model for each test trial using data that is dynamically selected from a candidate training set to match the conditions of the trial. In this work, we extend the TBC method, proposing: 1) a new similarity metric for selecting training data that result in significant gains over the one proposed in the original work; 2) a new option that enables the system to reject a trial when not enough matched data are available for training the calibration model; and 3) the use of regularization to improve the robustness of the calibration models trained for each trial. We test the proposed algorithms on a development set composed of several conditions and on the Federal Bureau of Investigation multi-condition speaker recognition dataset, and we demonstrate that the proposed approach reduces calibration loss to values close to 0 for most of the conditions when matched calibration data are available for selection, and that it can reject most of the trials for which relevant calibration data are unavailable.
publishDate 2019
dc.date.none.fl_str_mv 2019-01
dc.type.none.fl_str_mv info:eu-repo/semantics/article
info:eu-repo/semantics/publishedVersion
http://purl.org/coar/resource_type/c_6501
info:ar-repo/semantics/articulo
format article
status_str publishedVersion
dc.identifier.none.fl_str_mv http://hdl.handle.net/11336/123318
Ferrer, Luciana; Nandwana, Mahesh Kumar; McLaren, Mitchell; Castan, Diego; Lawson, Aaron; Toward Fail-Safe Speaker Recognition: Trial-Based Calibration with a Reject Option; Institute of Electrical and Electronics Engineers; IEEE/ACM Transactions on Audio Speech and Language Processing; 27; 1; 1-2019; 140-153
2329-9290
CONICET Digital
CONICET
url http://hdl.handle.net/11336/123318
identifier_str_mv Ferrer, Luciana; Nandwana, Mahesh Kumar; McLaren, Mitchell; Castan, Diego; Lawson, Aaron; Toward Fail-Safe Speaker Recognition: Trial-Based Calibration with a Reject Option; Institute of Electrical and Electronics Engineers; IEEE/ACM Transactions on Audio Speech and Language Processing; 27; 1; 1-2019; 140-153
2329-9290
CONICET Digital
CONICET
dc.language.none.fl_str_mv eng
language eng
dc.relation.none.fl_str_mv info:eu-repo/semantics/altIdentifier/url/https://ieeexplore.ieee.org/document/8490592
info:eu-repo/semantics/altIdentifier/doi/10.1109/TASLP.2018.2875794
dc.rights.none.fl_str_mv info:eu-repo/semantics/openAccess
https://creativecommons.org/licenses/by-nc-sa/2.5/ar/
eu_rights_str_mv openAccess
rights_invalid_str_mv https://creativecommons.org/licenses/by-nc-sa/2.5/ar/
dc.format.none.fl_str_mv application/pdf
application/pdf
dc.publisher.none.fl_str_mv Institute of Electrical and Electronics Engineers
publisher.none.fl_str_mv Institute of Electrical and Electronics Engineers
dc.source.none.fl_str_mv reponame:CONICET Digital (CONICET)
instname:Consejo Nacional de Investigaciones Científicas y Técnicas
reponame_str CONICET Digital (CONICET)
collection CONICET Digital (CONICET)
instname_str Consejo Nacional de Investigaciones Científicas y Técnicas
repository.name.fl_str_mv CONICET Digital (CONICET) - Consejo Nacional de Investigaciones Científicas y Técnicas
repository.mail.fl_str_mv dasensio@conicet.gov.ar; lcarlino@conicet.gov.ar
_version_ 1844614389833924608
score 13.070432