Speaker identification is deemed to be one of the basic tasks in audio forensics. Delivering a categorical opinion is often difficult due to insufficient quality of the recorded material, simulation or modulation of speaker’s voice. Hence, a wide-ranging approach to the identification process is used, including both subjective and objective methods. With their help, it becomes possible to obtain a broad spectrum of speech characteristics ranging from low-level features relating to physical construction of the vocal tract to advanced ones concerning various ways of expressing oneself and articulation, acquired during socialization process. This article describes an experiment undertaken to create acoustic profiles of a chosen group of speakers based on the features mentioned above.
Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member and would like to subscribe to the E-Library then Join the AES!
This paper costs $33 for non-members and is free for AES members and E-Library subscribers.