Speech Science Forum: Dr Harriet Smith
Title: Voice identification procedures: Improving earwitness performance. Room: 118, Chandler House

Voice identification procedures: Improving earwitness performance
In crimes when a perpetrator is heard but never never seen (e.g. masked attack or telephone fraud) an earwitness might be required to provide a description of the perpetrator’s voice, and then to attempt identification from a voice parade. Although such evidence can be pivotal, very little research has addressed a) the likely accuracy associated with these tasks, and b) whether existing procedures can be adapted in order to support optimal performance. In an ongoing program of research we are investigating methods of improving lay-listener voice descriptions in order to develop a procedure for eliciting accurate and admissible earwitness testimony. We are also addressing how identification accuracy and self-rated confidence vary across different parade procedures. The results I will present highlight the error-prone nature of earwitness performance, and underline the urgency of adapting current procedures in order to reduce the risk of miscarriages of justice.