Determination of Likelihood Ratios for Forensic Voice Comparison Using Principal Component Analysis

Authors

  • Balamurali Nair The University of Auckland
  • Esam Alzqhoul The University of Auckland
  • Bernard John Guillemin The University of Auckland

DOI:

https://doi.org/10.1558/ijsll.v21i1.83

Keywords:

Forensic voice comparison, Likelihood ratio, Multivariate kernel density, Condition number, Correlation, Principal component analysis

Abstract

The likelihood ratio (LR) framework is gaining increasing acceptance amongst forensic speech scientists when undertaking forensic voice comparison. Multivariate Kernel Density (MVKD) is one approach that has been used for calculating LRs when the number of parameters is in the region of 3 or 4. However there could be robustness issues with this approach when the number of parameters is larger than this. In this paper we present an alternative to the MVKD approach, termed Principal Component Analysis Kernel Density Likelihood Ratio (PCAKLR), which takes account of within-segment correlations, yet is computationally robust irrespective of the number of parameters used. We show that PCAKLR produces comparable results to MVKD for small numbers of parameters. Further, it also has the ability to directly handle between-segment correlations and is thus an alternative to the logistic-regression fusion typically used to combine results from multiple segments.

Author Biographies

Balamurali Nair, The University of Auckland

Balamurali Nair received his Bachelor of Technology degree from Kerala University, India, in 2007. He is currently a PhD student at the University of Auckland. His research study focuses on the impact of GSM mobile phone technology on forensic voice comparison.

Esam Alzqhoul, The University of Auckland

Esam Alzqhoul received his Bachelor of Electrical Engineering from the University of Jordan in 2006 and a Masters of Electrical & Electronic Engineering from the Near East University, Cyprus, in 2010. He is currently a PhD student at the University of Auckland, New Zealand. His research area focuses on the impact of CDMA mobile phone technology on forensic voice comparison.

Bernard John Guillemin, The University of Auckland

Bernard Guillemin received the PhD from the University of Auckland, New Zealand, in 1986. He is currently a Senior Lecturer in the Department of Electrical & Computer Engineering, University of Auckland. His research interests include forensic voice comparison, speech analysis, synthesis and recognition, speech enhancement, and active noise cancellation. He is a member of the International Association of Forensic Linguists (IAFL) as well as the Forensic Speech Science Committee within the Australasian Speech Science Technology Association (ASSTA).

Published

2014-06-26

How to Cite

Nair, B., Alzqhoul, E., & Guillemin, B. J. (2014). Determination of Likelihood Ratios for Forensic Voice Comparison Using Principal Component Analysis. International Journal of Speech, Language and the Law, 21(1), 83–112. https://doi.org/10.1558/ijsll.v21i1.83

Issue

Section

Articles