The effect of sample size on the performance of likelihood ratio based forensic voice comparison

Yuko Kinoshita, Shunichi Ishihara

Research output: A Conference proceeding or a Chapter in BookConference contribution

Abstract

This is a preliminary investigation into the effect of sample size on a likelihood ratio based forensic voice comparison system. We investigated how the number of vowel tokens used in the testing data affects LR estimation, using Japanese filler expressions elicited from spontaneous speech. In forensic voice comparison casework, we cannot control how many tokens of vowels or segments are available: sometimes only a few comparable tokens for each vowel can be found. We currently do not know if the LR produced using two tokens of a vowel is as reliable as one produced using ten tokens. We found that the two components of quality of LR — Cllr_min and Cllr_cal — responded very differently to additional tokens. Cllr_cal consistently deteriorated as tokens were added to the testing data, whereas Cllr_min improved rapidly. The improvement was particularly notable up to six tokens.
Original languageEnglish
Title of host publicationProceedings of the 14th Australasian International Conference on Speech Science and Technology
EditorsFelicity Cox, Katherine Demuth, Susan Lin, Kelly Miles, Sallyanne Palethrope, Jason Shaw, Ivan Yuen
Place of PublicationSydney
PublisherAustralian Speech Science and Technology Association (ASSTA)
Pages1-4
Number of pages4
Volume1
Publication statusPublished - 2012
Event14th Australasian International Conference on Speech Science and Technology - Sydney, Sydney, Australia
Duration: 3 Dec 20126 Dec 2012

Publication series

NameProceedings of the Australasian International Conference on Speech Science and Technology
PublisherAustralasian Speech Science and Technology Association (ASSTA)
Volume1
ISSN (Print)1039-0227

Conference

Conference14th Australasian International Conference on Speech Science and Technology
CountryAustralia
CitySydney
Period3/12/126/12/12

Fingerprint

Testing
Fillers

Cite this

Kinoshita, Y., & Ishihara, S. (2012). The effect of sample size on the performance of likelihood ratio based forensic voice comparison. In F. Cox, K. Demuth, S. Lin, K. Miles, S. Palethrope, J. Shaw, & I. Yuen (Eds.), Proceedings of the 14th Australasian International Conference on Speech Science and Technology (Vol. 1, pp. 1-4). (Proceedings of the Australasian International Conference on Speech Science and Technology; Vol. 1). Sydney: Australian Speech Science and Technology Association (ASSTA).
Kinoshita, Yuko ; Ishihara, Shunichi. / The effect of sample size on the performance of likelihood ratio based forensic voice comparison. Proceedings of the 14th Australasian International Conference on Speech Science and Technology. editor / Felicity Cox ; Katherine Demuth ; Susan Lin ; Kelly Miles ; Sallyanne Palethrope ; Jason Shaw ; Ivan Yuen. Vol. 1 Sydney : Australian Speech Science and Technology Association (ASSTA), 2012. pp. 1-4 (Proceedings of the Australasian International Conference on Speech Science and Technology).
@inproceedings{e6e73a805820444196871cfc21886c63,
title = "The effect of sample size on the performance of likelihood ratio based forensic voice comparison",
abstract = "This is a preliminary investigation into the effect of sample size on a likelihood ratio based forensic voice comparison system. We investigated how the number of vowel tokens used in the testing data affects LR estimation, using Japanese filler expressions elicited from spontaneous speech. In forensic voice comparison casework, we cannot control how many tokens of vowels or segments are available: sometimes only a few comparable tokens for each vowel can be found. We currently do not know if the LR produced using two tokens of a vowel is as reliable as one produced using ten tokens. We found that the two components of quality of LR — Cllr_min and Cllr_cal — responded very differently to additional tokens. Cllr_cal consistently deteriorated as tokens were added to the testing data, whereas Cllr_min improved rapidly. The improvement was particularly notable up to six tokens.",
keywords = "Forensic voice comparison, Likelihood ratio, Japanese",
author = "Yuko Kinoshita and Shunichi Ishihara",
year = "2012",
language = "English",
volume = "1",
series = "Proceedings of the Australasian International Conference on Speech Science and Technology",
publisher = "Australian Speech Science and Technology Association (ASSTA)",
pages = "1--4",
editor = "Felicity Cox and Katherine Demuth and Susan Lin and Kelly Miles and Sallyanne Palethrope and Jason Shaw and Ivan Yuen",
booktitle = "Proceedings of the 14th Australasian International Conference on Speech Science and Technology",

}

Kinoshita, Y & Ishihara, S 2012, The effect of sample size on the performance of likelihood ratio based forensic voice comparison. in F Cox, K Demuth, S Lin, K Miles, S Palethrope, J Shaw & I Yuen (eds), Proceedings of the 14th Australasian International Conference on Speech Science and Technology. vol. 1, Proceedings of the Australasian International Conference on Speech Science and Technology, vol. 1, Australian Speech Science and Technology Association (ASSTA), Sydney, pp. 1-4, 14th Australasian International Conference on Speech Science and Technology, Sydney, Australia, 3/12/12.

The effect of sample size on the performance of likelihood ratio based forensic voice comparison. / Kinoshita, Yuko; Ishihara, Shunichi.

Proceedings of the 14th Australasian International Conference on Speech Science and Technology. ed. / Felicity Cox; Katherine Demuth; Susan Lin; Kelly Miles; Sallyanne Palethrope; Jason Shaw; Ivan Yuen. Vol. 1 Sydney : Australian Speech Science and Technology Association (ASSTA), 2012. p. 1-4 (Proceedings of the Australasian International Conference on Speech Science and Technology; Vol. 1).

Research output: A Conference proceeding or a Chapter in BookConference contribution

TY - GEN

T1 - The effect of sample size on the performance of likelihood ratio based forensic voice comparison

AU - Kinoshita, Yuko

AU - Ishihara, Shunichi

PY - 2012

Y1 - 2012

N2 - This is a preliminary investigation into the effect of sample size on a likelihood ratio based forensic voice comparison system. We investigated how the number of vowel tokens used in the testing data affects LR estimation, using Japanese filler expressions elicited from spontaneous speech. In forensic voice comparison casework, we cannot control how many tokens of vowels or segments are available: sometimes only a few comparable tokens for each vowel can be found. We currently do not know if the LR produced using two tokens of a vowel is as reliable as one produced using ten tokens. We found that the two components of quality of LR — Cllr_min and Cllr_cal — responded very differently to additional tokens. Cllr_cal consistently deteriorated as tokens were added to the testing data, whereas Cllr_min improved rapidly. The improvement was particularly notable up to six tokens.

AB - This is a preliminary investigation into the effect of sample size on a likelihood ratio based forensic voice comparison system. We investigated how the number of vowel tokens used in the testing data affects LR estimation, using Japanese filler expressions elicited from spontaneous speech. In forensic voice comparison casework, we cannot control how many tokens of vowels or segments are available: sometimes only a few comparable tokens for each vowel can be found. We currently do not know if the LR produced using two tokens of a vowel is as reliable as one produced using ten tokens. We found that the two components of quality of LR — Cllr_min and Cllr_cal — responded very differently to additional tokens. Cllr_cal consistently deteriorated as tokens were added to the testing data, whereas Cllr_min improved rapidly. The improvement was particularly notable up to six tokens.

KW - Forensic voice comparison

KW - Likelihood ratio

KW - Japanese

M3 - Conference contribution

VL - 1

T3 - Proceedings of the Australasian International Conference on Speech Science and Technology

SP - 1

EP - 4

BT - Proceedings of the 14th Australasian International Conference on Speech Science and Technology

A2 - Cox, Felicity

A2 - Demuth, Katherine

A2 - Lin, Susan

A2 - Miles, Kelly

A2 - Palethrope, Sallyanne

A2 - Shaw, Jason

A2 - Yuen, Ivan

PB - Australian Speech Science and Technology Association (ASSTA)

CY - Sydney

ER -

Kinoshita Y, Ishihara S. The effect of sample size on the performance of likelihood ratio based forensic voice comparison. In Cox F, Demuth K, Lin S, Miles K, Palethrope S, Shaw J, Yuen I, editors, Proceedings of the 14th Australasian International Conference on Speech Science and Technology. Vol. 1. Sydney: Australian Speech Science and Technology Association (ASSTA). 2012. p. 1-4. (Proceedings of the Australasian International Conference on Speech Science and Technology).