Speaker Identification Problem (From Advances in Forensic Psychology and Psychiatry, P 87-111, 1984)

• NCJ Number: 122906
• Author(s): H Hollien; B Klepper
• Date Published: 1984
• Length: 25 pages
• Sale Source: Ablex Publishing Corporation
• Publication Type: Report (Technical)
• Language: English
• Country: United States of America

Annotation

This analysis of three approaches to the identification of speakers from analysis of their voice and speech alone emphasizes that no reliable method has yet been developed that will provide for the positive identification of individuals based on forensic standards.

Abstract

The three methods are the aural/perceptual approach, the use of voiceprints or voicegrams, and the machine approach. Of these, only the machine approach shows potential for developing an effective system. One such system has been developed by the IASCP group. This Semi-Automatic Speaker Identification system (SAUSI) focuses on five factors: 1) long-term speech spectra, 2) speaking fundamental frequency, 3) vowel formant tracking, 4) temporal analysis, and 5) intensity analysis. More research and experimentation is needed before such systems are made available for general use, however. 101 references.