Examining long-term formant distributions as a discriminant in forensic speaker comparisons under a likelihood ratio framework

Erica Gold, Peter French, Philip Harrison

Research output: Contribution to journalArticle

3 Citations (Scopus)

Abstract

This study investigates the use of long-term formant distributions (LTFDs) as a discriminant in forensic speaker comparisons. LTFDs are the distributions calculated for all values of each formant for a speaker in a single recording. Spontaneous speech recordings from 100 male speakers of Southern Standard British English were analyzed from the DyViS Database (Nolan et al. 2009). The recordings were auto-segmented to obtain a minimum of 50 seconds of vowels per speaker. The iCAbS (iterative cepstral analysis by synthesis) formant tracker was used to automatically extract and measure F1-F4. To assess the evidential value of the LTFDs, likelihood ratios (LRs) were computed using a Multivariate Kernel-Density approach (MVKD). It was found that LTFD performs well overall, but much better with different speaker comparisons than same speaker comparisons (97.4% compared to 84% of comparisons providing correct support). LTFD appears to be a good discriminant to include in forensic speaker comparison analyses and offers the added benefit of avoiding potential correlation problems between vowel phonemes. The MVKD based LR results from this study were also found to be comparable to those results in French et al. (2012) and Becker et al. (2008), which used a Gaussian Mixture Model-Universal Background Model LR approach.

Original languageEnglish
Article number060041
Number of pages8
JournalProceedings of Meetings on Acoustics
Volume19
Early online dateMay 2013
DOIs
Publication statusPublished - 2013
Externally publishedYes

Fingerprint

likelihood ratio
vowels
recording
cepstral analysis
phonemes
synthesis

Cite this

@article{9ee0e8f2f1794ea3b7d338aa0776ccf6,
title = "Examining long-term formant distributions as a discriminant in forensic speaker comparisons under a likelihood ratio framework",
abstract = "This study investigates the use of long-term formant distributions (LTFDs) as a discriminant in forensic speaker comparisons. LTFDs are the distributions calculated for all values of each formant for a speaker in a single recording. Spontaneous speech recordings from 100 male speakers of Southern Standard British English were analyzed from the DyViS Database (Nolan et al. 2009). The recordings were auto-segmented to obtain a minimum of 50 seconds of vowels per speaker. The iCAbS (iterative cepstral analysis by synthesis) formant tracker was used to automatically extract and measure F1-F4. To assess the evidential value of the LTFDs, likelihood ratios (LRs) were computed using a Multivariate Kernel-Density approach (MVKD). It was found that LTFD performs well overall, but much better with different speaker comparisons than same speaker comparisons (97.4{\%} compared to 84{\%} of comparisons providing correct support). LTFD appears to be a good discriminant to include in forensic speaker comparison analyses and offers the added benefit of avoiding potential correlation problems between vowel phonemes. The MVKD based LR results from this study were also found to be comparable to those results in French et al. (2012) and Becker et al. (2008), which used a Gaussian Mixture Model-Universal Background Model LR approach.",
keywords = "Speech analysis, Forensic science, Phonetic segments, Databases, Data analysis",
author = "Erica Gold and Peter French and Philip Harrison",
year = "2013",
doi = "10.1121/1.4800285",
language = "English",
volume = "19",
journal = "Proceedings of Meetings on Acoustics",
issn = "1939-800X",
publisher = "Acoustical Society of America",

}

Examining long-term formant distributions as a discriminant in forensic speaker comparisons under a likelihood ratio framework. / Gold, Erica; French, Peter; Harrison, Philip.

In: Proceedings of Meetings on Acoustics, Vol. 19, 060041, 2013.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Examining long-term formant distributions as a discriminant in forensic speaker comparisons under a likelihood ratio framework

AU - Gold, Erica

AU - French, Peter

AU - Harrison, Philip

PY - 2013

Y1 - 2013

N2 - This study investigates the use of long-term formant distributions (LTFDs) as a discriminant in forensic speaker comparisons. LTFDs are the distributions calculated for all values of each formant for a speaker in a single recording. Spontaneous speech recordings from 100 male speakers of Southern Standard British English were analyzed from the DyViS Database (Nolan et al. 2009). The recordings were auto-segmented to obtain a minimum of 50 seconds of vowels per speaker. The iCAbS (iterative cepstral analysis by synthesis) formant tracker was used to automatically extract and measure F1-F4. To assess the evidential value of the LTFDs, likelihood ratios (LRs) were computed using a Multivariate Kernel-Density approach (MVKD). It was found that LTFD performs well overall, but much better with different speaker comparisons than same speaker comparisons (97.4% compared to 84% of comparisons providing correct support). LTFD appears to be a good discriminant to include in forensic speaker comparison analyses and offers the added benefit of avoiding potential correlation problems between vowel phonemes. The MVKD based LR results from this study were also found to be comparable to those results in French et al. (2012) and Becker et al. (2008), which used a Gaussian Mixture Model-Universal Background Model LR approach.

AB - This study investigates the use of long-term formant distributions (LTFDs) as a discriminant in forensic speaker comparisons. LTFDs are the distributions calculated for all values of each formant for a speaker in a single recording. Spontaneous speech recordings from 100 male speakers of Southern Standard British English were analyzed from the DyViS Database (Nolan et al. 2009). The recordings were auto-segmented to obtain a minimum of 50 seconds of vowels per speaker. The iCAbS (iterative cepstral analysis by synthesis) formant tracker was used to automatically extract and measure F1-F4. To assess the evidential value of the LTFDs, likelihood ratios (LRs) were computed using a Multivariate Kernel-Density approach (MVKD). It was found that LTFD performs well overall, but much better with different speaker comparisons than same speaker comparisons (97.4% compared to 84% of comparisons providing correct support). LTFD appears to be a good discriminant to include in forensic speaker comparison analyses and offers the added benefit of avoiding potential correlation problems between vowel phonemes. The MVKD based LR results from this study were also found to be comparable to those results in French et al. (2012) and Becker et al. (2008), which used a Gaussian Mixture Model-Universal Background Model LR approach.

KW - Speech analysis

KW - Forensic science

KW - Phonetic segments

KW - Databases

KW - Data analysis

UR - http://www.scopus.com/inward/record.url?scp=84878972827&partnerID=8YFLogxK

UR - http://asa.scitation.org/journal/pma

U2 - 10.1121/1.4800285

DO - 10.1121/1.4800285

M3 - Article

VL - 19

JO - Proceedings of Meetings on Acoustics

JF - Proceedings of Meetings on Acoustics

SN - 1939-800X

M1 - 060041

ER -