Text-independent Forensic Speaker Identification Using Telephone Speech

On several occasions we have been asked by police officials whether it is feasible for investigative purposes to automatically identify a speaker using speech from intercepted telephone conversations. Since forensic phoneticians are typically not involved in speaker identification but in speaker verification we tend to reformulate this question in : Is it feasible for investigative purposes to automatically verify a speaker’s identity using a database built from interceped telephone conversations? This paper presents an experiment in which a closed set text independent speaker identification method is tested on telephone speech. The speech database used was built from telephone speech recorded for investigative purposes. The distance measure used for the speaker recognition is related to second order statistical tests. It is expressed as a function of the eigenvalues of a test covariance matrix relative to a training covariance matrix. The text independent recognition technique is based on distance measures developed by Bimbot and Mathan in 1995 [l].