This project will establish links between the shape of an individual’s vocal tract and identifying characteristics in their speech, based on detailed anatomical information from magnetic resonance images (MRI) and established acoustic modelling techniques. The impact of anatomical differences upon speech is of particular interest in forensic speaker comparison (FSC), as speech events are unique even when repeated under identical circumstances, let alone the different circumstances usually encountered in forensic casework. Anatomy, on the other hand, is invariant for a speaker, potentially providing acoustic cues for identification across recordings. This project asks three research questions relating to anatomical differences: which parts of the vocal anatomy have the most impact upon speaker-discriminatory features in the speech signal?; what degree of anatomical variation can be expected among the population?; and to what extent are these differences detectable by naive listeners, FSC experts, and automatic speaker recognition (ASR) systems? The applicant has unique insight into both acoustic engineering methods and FSC, and this project will use a novel combination of techniques from both disciplines to address the problem at a level well beyond the simplified models used in previous acoustic-anatomical studies. The increasing availability of MRI vocal tract data allows the impact of anatomy on speech to be studied at a much larger scale and in much greater detail than has previously been possible. The project will be developed alongside FSC experts to ensure that the results are applicable to forensic casework from the outset. The results have further applications in speaker verification, oral surgery planning, speech and language therapy, and personalised speech synthesis.