Quality measures based calibration with duration and noise dependency for speaker recognition

Miranti Indar Mandasari; Rahim Saeidi; David A. Van Leeuwen

doi:10.1016/j.specom.2015.05.009

Quality measures based calibration with duration and noise dependency for speaker recognition

Miranti Indar Mandasari, Rahim Saeidi^*, David A. Van Leeuwen

^*Corresponding author for this work

Department of Signal Processing and Acoustics

Research output: Contribution to journal › Article › Scientific › peer-review

17 Citations (Scopus)

Abstract

Abstract This paper studies the effect of short utterances and noise on the performance of automatic speaker recognition. We focus on calibration aspects, and propose a calibration strategy that uses quality measures to model the calibration parameters. We carry out the proposed calibration by using simple Quality Measure Functions (QMFs) of duration and measured signal-to-noise-ratio from speech segments. We test the effectiveness of the approach using two databases, the development set of the I4U collaboration for the NIST Speaker Recognition Evaluation (SRE) 2012, and the evaluation test material of NIST SRE 2012 itself. In comparison with conventional linear calibration, results show that the proposed QMF approach successfully improves the system performance in terms of both discrimination and calibration.

Original language	English
Article number	2297
Pages (from-to)	126-137
Number of pages	12
Journal	Speech Communication
Volume	72
DOIs	https://doi.org/10.1016/j.specom.2015.05.009
Publication status	Published - 14 Jun 2015
MoE publication type	A1 Journal article-refereed

Keywords

Calibration
Duration
Noise
Quality measure function
Speaker recognition

Access to Document

10.1016/j.specom.2015.05.009

Link to publication in Scopus

Fingerprint Dive into the research topics of 'Quality measures based calibration with duration and noise dependency for speaker recognition'. Together they form a unique fingerprint.

Cite this

@article{8ed4580d3e8f423a913889d72c1dbbba,

title = "Quality measures based calibration with duration and noise dependency for speaker recognition",

abstract = "Abstract This paper studies the effect of short utterances and noise on the performance of automatic speaker recognition. We focus on calibration aspects, and propose a calibration strategy that uses quality measures to model the calibration parameters. We carry out the proposed calibration by using simple Quality Measure Functions (QMFs) of duration and measured signal-to-noise-ratio from speech segments. We test the effectiveness of the approach using two databases, the development set of the I4U collaboration for the NIST Speaker Recognition Evaluation (SRE) 2012, and the evaluation test material of NIST SRE 2012 itself. In comparison with conventional linear calibration, results show that the proposed QMF approach successfully improves the system performance in terms of both discrimination and calibration.",

keywords = "Calibration, Duration, Noise, Quality measure function, Speaker recognition",

author = "Mandasari, {Miranti Indar} and Rahim Saeidi and {Van Leeuwen}, {David A.}",

year = "2015",

month = jun,

day = "14",

doi = "10.1016/j.specom.2015.05.009",

language = "English",

volume = "72",

pages = "126--137",

journal = "Speech Communication",

issn = "0167-6393",

publisher = "Elsevier",

}

TY - JOUR

T1 - Quality measures based calibration with duration and noise dependency for speaker recognition

AU - Mandasari, Miranti Indar

AU - Saeidi, Rahim

AU - Van Leeuwen, David A.

PY - 2015/6/14

Y1 - 2015/6/14

N2 - Abstract This paper studies the effect of short utterances and noise on the performance of automatic speaker recognition. We focus on calibration aspects, and propose a calibration strategy that uses quality measures to model the calibration parameters. We carry out the proposed calibration by using simple Quality Measure Functions (QMFs) of duration and measured signal-to-noise-ratio from speech segments. We test the effectiveness of the approach using two databases, the development set of the I4U collaboration for the NIST Speaker Recognition Evaluation (SRE) 2012, and the evaluation test material of NIST SRE 2012 itself. In comparison with conventional linear calibration, results show that the proposed QMF approach successfully improves the system performance in terms of both discrimination and calibration.

AB - Abstract This paper studies the effect of short utterances and noise on the performance of automatic speaker recognition. We focus on calibration aspects, and propose a calibration strategy that uses quality measures to model the calibration parameters. We carry out the proposed calibration by using simple Quality Measure Functions (QMFs) of duration and measured signal-to-noise-ratio from speech segments. We test the effectiveness of the approach using two databases, the development set of the I4U collaboration for the NIST Speaker Recognition Evaluation (SRE) 2012, and the evaluation test material of NIST SRE 2012 itself. In comparison with conventional linear calibration, results show that the proposed QMF approach successfully improves the system performance in terms of both discrimination and calibration.

KW - Calibration

KW - Duration

KW - Noise

KW - Quality measure function

KW - Speaker recognition

UR - http://www.scopus.com/inward/record.url?scp=84930948085&partnerID=8YFLogxK

U2 - 10.1016/j.specom.2015.05.009

DO - 10.1016/j.specom.2015.05.009

M3 - Article

AN - SCOPUS:84930948085

VL - 72

SP - 126

EP - 137

JO - Speech Communication

JF - Speech Communication

SN - 0167-6393

M1 - 2297

ER -