Front-end approaches to the issue of correlations in forensic speaker comparison

Erica Gold, Vincent Hughes

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

In likelihood ratio (LR)-based forensic speaker comparison it is essential to consider correlations between parameters to accurately estimate the overall strength of the evidence. Current approaches attempt to deal with correlations after the computation of LRs (back-end processing). This paper explores alternative, front-end techniques, which consider the underlying correlation structure of the raw data. Calibrated LRs were computed for a range of parameters commonly analysed in speaker comparisons. LRs were combined using (1) an assumption of independence, (2) the mean, (3) assumptions from phonetic theory, and (4) empirical correlations in the raw data. System (1), based on an assumption of independence, produced the best validity (Cllr = 0.04). Predictably, overall strength of evidence was also highest for system (1), while strength of evidence was weakest using the mean (2). Both systems (3) and (4) performed well achieving Cllr values of ca. 0.09.
LanguageEnglish
Title of host publicationProceedings of the 18th International Congress of Phonetic Sciences
Place of PublicationGlasgow, UK
PublisherUniversity of Glasgow
Pages1-5
Number of pages5
ISBN (Print)9780852619414
StatePublished - 2015
Event18th International Congress of Phonetic Sciences - Scottish Event Campus, Glasgow, United Kingdom
Duration: 10 Aug 201514 Aug 2015
Conference number: 18
https://www.internationalphoneticassociation.org/icphs/icphs2015 (Link to Conference Details)

Conference

Conference18th International Congress of Phonetic Sciences
Abbreviated titleICPhS 2015
CountryUnited Kingdom
CityGlasgow
Period10/08/1514/08/15
Internet address

Fingerprint

comparison
parameter

Cite this

Gold, E., & Hughes, V. (2015). Front-end approaches to the issue of correlations in forensic speaker comparison. In Proceedings of the 18th International Congress of Phonetic Sciences (pp. 1-5). [486] Glasgow, UK: University of Glasgow.
Gold, Erica ; Hughes, Vincent. / Front-end approaches to the issue of correlations in forensic speaker comparison. Proceedings of the 18th International Congress of Phonetic Sciences. Glasgow, UK : University of Glasgow, 2015. pp. 1-5
@inproceedings{c8bcb35a66124287a4874bf02073ffad,
title = "Front-end approaches to the issue of correlations in forensic speaker comparison",
abstract = "In likelihood ratio (LR)-based forensic speaker comparison it is essential to consider correlations between parameters to accurately estimate the overall strength of the evidence. Current approaches attempt to deal with correlations after the computation of LRs (back-end processing). This paper explores alternative, front-end techniques, which consider the underlying correlation structure of the raw data. Calibrated LRs were computed for a range of parameters commonly analysed in speaker comparisons. LRs were combined using (1) an assumption of independence, (2) the mean, (3) assumptions from phonetic theory, and (4) empirical correlations in the raw data. System (1), based on an assumption of independence, produced the best validity (Cllr = 0.04). Predictably, overall strength of evidence was also highest for system (1), while strength of evidence was weakest using the mean (2). Both systems (3) and (4) performed well achieving Cllr values of ca. 0.09.",
keywords = "Likelihood ratio, Forensic speaker comparison, Correlations, Front-end processing",
author = "Erica Gold and Vincent Hughes",
year = "2015",
language = "English",
isbn = "9780852619414",
pages = "1--5",
booktitle = "Proceedings of the 18th International Congress of Phonetic Sciences",
publisher = "University of Glasgow",

}

Gold, E & Hughes, V 2015, Front-end approaches to the issue of correlations in forensic speaker comparison. in Proceedings of the 18th International Congress of Phonetic Sciences., 486, University of Glasgow, Glasgow, UK, pp. 1-5, 18th International Congress of Phonetic Sciences, Glasgow, United Kingdom, 10/08/15.

Front-end approaches to the issue of correlations in forensic speaker comparison. / Gold, Erica; Hughes, Vincent.

Proceedings of the 18th International Congress of Phonetic Sciences. Glasgow, UK : University of Glasgow, 2015. p. 1-5 486.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Front-end approaches to the issue of correlations in forensic speaker comparison

AU - Gold,Erica

AU - Hughes,Vincent

PY - 2015

Y1 - 2015

N2 - In likelihood ratio (LR)-based forensic speaker comparison it is essential to consider correlations between parameters to accurately estimate the overall strength of the evidence. Current approaches attempt to deal with correlations after the computation of LRs (back-end processing). This paper explores alternative, front-end techniques, which consider the underlying correlation structure of the raw data. Calibrated LRs were computed for a range of parameters commonly analysed in speaker comparisons. LRs were combined using (1) an assumption of independence, (2) the mean, (3) assumptions from phonetic theory, and (4) empirical correlations in the raw data. System (1), based on an assumption of independence, produced the best validity (Cllr = 0.04). Predictably, overall strength of evidence was also highest for system (1), while strength of evidence was weakest using the mean (2). Both systems (3) and (4) performed well achieving Cllr values of ca. 0.09.

AB - In likelihood ratio (LR)-based forensic speaker comparison it is essential to consider correlations between parameters to accurately estimate the overall strength of the evidence. Current approaches attempt to deal with correlations after the computation of LRs (back-end processing). This paper explores alternative, front-end techniques, which consider the underlying correlation structure of the raw data. Calibrated LRs were computed for a range of parameters commonly analysed in speaker comparisons. LRs were combined using (1) an assumption of independence, (2) the mean, (3) assumptions from phonetic theory, and (4) empirical correlations in the raw data. System (1), based on an assumption of independence, produced the best validity (Cllr = 0.04). Predictably, overall strength of evidence was also highest for system (1), while strength of evidence was weakest using the mean (2). Both systems (3) and (4) performed well achieving Cllr values of ca. 0.09.

KW - Likelihood ratio

KW - Forensic speaker comparison

KW - Correlations

KW - Front-end processing

UR - https://www.internationalphoneticassociation.org/icphs/icphs2015

M3 - Conference contribution

SN - 9780852619414

SP - 1

EP - 5

BT - Proceedings of the 18th International Congress of Phonetic Sciences

PB - University of Glasgow

CY - Glasgow, UK

ER -

Gold E, Hughes V. Front-end approaches to the issue of correlations in forensic speaker comparison. In Proceedings of the 18th International Congress of Phonetic Sciences. Glasgow, UK: University of Glasgow. 2015. p. 1-5. 486.