Likelihood ratio-based forensic voice comparison with higher level features: research and reality. (September 2017)

• Record Type: Journal Article
• Authors: Rose, Phil
• Is Part Of: Computer speech & language. Volume 45(2017)
• Page Start: 475
• Page End: 502
• Publication Date: 2017-09
• Subjects: Forensic voice comparison -- Likelihood ratio -- Cantonese -- Higher-level features -- F-pattern trajectories -- Tonal F0 trajectory -- Short term F0 -- Segmental cepstrum; Speech processing systems -- Periodicals; Automatic speech recognition -- Periodicals; Computers -- Periodicals; Linguistics -- Periodicals; Speech-Language Pathology -- Periodicals; Traitement automatique de la parole -- Périodiques; Reconnaissance automatique de la parole -- Périodiques; Automatic speech recognition; Speech processing systems; Electronic journals; Periodicals; 006.454
• DOI: 10.1016/j.csl.2017.03.003
• Languages: English
• ISSNs: 0885-2308

Abstract

Highlights: Examples are given of the legally and logically correct way of estimating the strength of forensic voice comparison evidence with likelihood ratios both in real world cases and research. It is shown how higher level features have helped in real world forensic case work involving serious crimes of a $150 million telephone fraud, ice importation, match fixing and a terrorist bomb hoax phone call. It is shown with Cantonese data how higher level features can be combined to give greater strength of forensic speaker recognition evidence. Important pros and cons of likelihood ratios are discussed. Important pros and cons of higher level features for forensic voice comparison are discussed and exemplified. Abstract: Examples are given of forensic voice comparison with higher level features in real-world cases and research. A pilot experiment relating to estimation of strength of evidence in forensic voice comparison is described which explores the use of higher-level features extracted over a disyllabic word as a whole, rather than over individual monosyllables as conventionally practiced. The trajectories of the first three formants and tonal F0 of the hexaphonic disyllabic Cantonese word daihyat 'first' from controlled but natural non-contemporaneous recordings of 23 male speakers are modeled with polynomials, and multivariate likelihood ratios estimated from their coefficients. Evaluation with the log likelihood ratio cost validity metric C llr shows an optimum performance is obtained, surprisingly, with lower order polynomials, with F2 requiring a cubic fit, and F1 and F3 quadratic. Fusion of F-pattern and tonal F0 results in considerable improvement over the individual features, reducing the C llr to ca. 0.1. The forensic potential of the daihyat data is demonstrated by fusion with three other Cantonese higher-level features: the F-pattern of /i/, short-term F0, and syllabic nasal cepstral spectrum, which reduces the C llr still further to 0.03. Important pros and cons of higher-level features and likelihood ratios are discussed, the latter illustrated with data from Japanese, and three varieties of English in real forensic casework.