The success of automatic speaker recognition in laboratory environments suggests applications in forensic science for establishing the identity of individuals on the basis of features extracted from speech. A theoretical model for such a verification scheme for continuous normally distributed features is developed. The three cases of using a) single feature, b) multiple independent measurements of a single feature, and c) multiple independent features are explored. The number of independent features needed for a reliable personal identification is computed based on the theoretical model and an exploratory study of some speech features.

Abstract

1980