How reliably can individuals be recognised by their voices? This question has recently been the subject of much debate among speech researchers and forensic scientists and the controversial and crucial nature of that debate has stimulated a wide range of empirical research. In this book Dr Nolan argues convincingly that both the design and interpretation of many of these experiments are vitiated by the lack of a comprehensive model of variability between speakers and within the speech of an individual. This volume clearly demonstrates that any valid theory of speaker recognition must integrate the approaches of a number of disciplines and it is itself an important step towards that integration. It will be of interest to phoneticians and to speech scientists, including those with an engineering background and also to forensic scientists specialising in this area.