Where the science ends and the law begins: Likelihood ratio-based forensic voice comparison in a $150 million telephone fraud

The first use of likelihood ratios for the evaluation of forensic voice comparison evidence in a real trial in Australia is documented. Important steps in the process of estimating the strength of evidence - from acoustic-phonetic features in the utterances 'yes' and 'not too bad' - are described and explained. These comprise the nature and currency of likelihood ratios, choice of features, graphical demonstration of similarity and typicality of suspect and offender speech samples, and their multivariate likelihood ratio evaluation against a reference sample. Improvements in likelihood ratio estimation since the case are exemplified; in particular how to demonstrate that the approach actually works. The reception of the likelihood ratio-based evidence during the trial is used to highlight some problems in its rational evaluation by a court.