Forensic speaker recognition based on a Bayesian framework and Gaussian mixture modelling (GMM)

The goal of this paper is to establish a scientifically founded methodology for forensic automatic speaker recognition. The interpretation of recorded speech as evidence in the forensic context presents particular challenges. The means proposed in the paper for dealing with them is through Bayesian inference. This leads to the formulation of a likelihood ratio measure of evidence which weighs the evidence in favor of two competing hypotheses: 1) the suspected speaker is the source of the questioned recording (trace), 2) the speaker at the origin of the questioned recording is not the suspected speaker. The state-of-the-art automatic recognition system using Gaussian mixture model (GMM) is adapted to the Bayesian interpretation (BI) framework with the models of the within-source variability of the suspected speaker and the between-source variability of the questioned recording. This double-statistical approach (BI-GMM) gives an adequate solution for the interpretation of the recorded speech as evidence in the judicial process. Examples provided are for telephone quality speech recordings that account for a very large proportion of all forensic material for speaker recognition.