Evidential value of voice quality acoustics in forensic voice comparison (Research assistant)

2019 - 2020 (Early Career Scheme, Hong Kong Research Grants Council, HKD 450,000)

with Ricky Chan (PI, University of Hong Kong)

The task of forensic voice comparison typically involves the comparison of a known voice of the suspect with an unknown voice of the offender. In cases which involve speech recordings (e.g. hoax calls, ransom demands, threatening voice message, conversation with an accomplice), the comparison of voices can help the trier-of-fact (e.g. judge/jury) or investigating authorities (e.g. police) decide whether the two voices were produced by the same speaker or different individuals. A major goal of forensic voice comparison research is to identify features in speech which are useful for distinguishing voices. Voice quality (e.g. breathy/creaky/hoarse/nasalised voice) analysis by phoneticians has been reported to be useful for characterising speakers and widely conducted in forensic casework, but voice quality as forensic evidence has mostly been assessed subjectively by experts and reported in terms of categorical auditory labels. Across different fields of forensic comparison sciences, there has been increasing pressure from regulating bodies (e.g. US National Research Council and UK Law Commission) for forensic experts to shift from relying on subjective judgments to the quantitative assessment of the strength of the evidence using the likelihood ratio framework. This project explores the quantitative aspects of voice quality and involves likelihood-ratio analyses of voice quality acoustics in a bid to determine their evidential value.