Forensic voice comparison with monophthongal formant trajectories - a likelihood ratio-based discrimination of "schwa" vowel acoustics in a close social group of young Australian females

Abstract

An experiment is described relating to estimation of strength of evidence in likelihood ratio-based forensic voice comparison. It is asked whether a better performance is obtained from point estimation of formant pattern targets in monophthongal vowel acoustics rather than formant trajectories. The hypothesis is tested on non-contemporaneous recordings of a custom-built challenging database of 26 young Australian female voices performing a map task. Evaluation with the log likelihood ratio cost validity metric Cllr shows that both trajectory and target perform well, but that contrary to phonological predictions, evidence based on monophthongal F-pattern trajectory is superior to target point measurements.