skip navigation

PUBLICATIONS

Register for Latest Research

Stay Informed
Register with NCJRS to receive NCJRS's biweekly e-newsletter JUSTINFO and additional periodic emails from NCJRS and the NCJRS federal sponsors that highlight the latest research published or sponsored by the Office of Justice Programs.

NCJRS Abstract

The document referenced below is part of the NCJRS Virtual Library collection. To conduct further searches of the collection, visit the Virtual Library. See the Obtain Documents page for direction on how to access resources online, via mail, through interlibrary loans, or in a local library.

 

NCJ Number: 31260 Find in a Library
Title: DEVELOPMENT OF ANALYTICAL METHODS FOR A SEMI-AUTOMATIC SPEAKER IDENTIFICATION SYSTEM (FROM CARNAHAN CONFERENCE ON CRIME COUNTERMEASURES, 1975 - PROCEEDINGS BY JOHN S JACKSON - SEE NCJ-31252)
Author(s): J E PAUL; A S RABINOWITZ; J P RIGANATI; J M RICHARDSON
Corporate Author: University of Kentucky
College of Engineering
United States of America

Institute of Electrical and Electronics Engineers
Lexington Section
United States of America

Aerospace and Electronic Systems Society
United States of America
Date Published: 1975
Page Count: 13
Sponsoring Agency: Aerospace and Electronic Systems Society

Institute of Electrical and Electronics Engineers
Lexington, KY
University of Kentucky
Lexington, KY 40506
Type: Report (Study/Research)
Format: Document
Language: English
Country: United States of America
Annotation: THIS PAPER DESCRIBES THE ANALYTICAL STUDIES LEADING UP TO THE DEVELOPMENT OF SASIS (SEMI-AUTOMATIC SPEAKER IDENTIFICATION SYSTEM) WHICH UTILIZES COMPUTERS TO ANALYZE AND COMPARE SEGMENTS OF SPEECH.
Abstract: IN THE SASIS, SEGMENTS OF SPEECH ARE EXTRACTED FROM TWO SPEECH UTTERANCES AND ARE COMPUTER-ANALYZED TO YIELD A STATISTICAL MEASURE INDICATING WHETHER THE UTTERANCES WERE SAID BY THE SAME OR DIFFERENT SPEAKERS. THE ANALYTICAL STUDIES INVOLVED THE COLLECTION AND PROCESSING OF SPEECH DATA FROM OVER 250 SPEAKERS FROM WHICH WAS EXTRACTED OVER 35,000 PHONETIC-EVENT TOKENS. A STUDY OF THE DISCRIMINATING POWER OF INDIVIDUAL PHONETIC EVENTS, ALONG WITH A STUDY OF COARTICULATION EFFECTS ON THESE SAME EVENTS, WAS CONDUCTED. A UNIQUE SET OF 30 FEATURES WAS DEFINED FOR EACH OF THE RESULTANT 13 PHONETIC CATEGORIES USED BY THE SASIS IN CARRYING OUT SPEAKER IDENTIFICATION. A WEIGHTED EUCLIDEAN DISTANCE MEASURE WAS SELECTED FOR MEASURING SPEAKER SIMILARITY WITHIN A PHONETIC CATEGORY, AND A DESENSITIZED FISHER DISCRIMINANT WAS INCORPORATED TO COMBINE THESE INDIVIDUAL DISTANCE MEASURES INTO AN OVERALL MEASURE OF SIMILARITY BETWEEN THE SPEAKER OF TWO SPOKEN UTTERANCES. A SET OF LIKELIHOOD TABLES, INDEXED BY SIMILARITY MEASURE AND THE PHONETIC EVENTS USED, WAS DERIVED TO DETERMINE THE PROBABILITY THAT THE UTTERANCES WERE UTTERED BY THE SAME OR DIFFERENT INDIVIDUALS. (AUTHOR ABSTRACT MODIFIED)
Index Term(s): Automated police information systems; Computer aided operations; Minicomputer systems; Research; Research and development; Suspect identification; Voice identification
To cite this abstract, use the following link:
http://www.ncjrs.gov/App/publications/abstract.aspx?ID=31260

*A link to the full-text document is provided whenever possible. For documents not available online, a link to the publisher's website is provided. Tell us how you use the NCJRS Library and Abstracts Database - send us your feedback.