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: 66275 Find in a Library
Title: COMPUTER-ASSISTED VOICE IDENTIFICATION - WEST GERMANY
Journal: KRIMINALISTIK  Volume:31  Dated:(1977)  Pages:529-540
Author(s): E BUNGE
Corporate Author: Kriminalistik Verlag Gmbh
Germany (Unified)
Date Published: 1978
Page Count: 11
Sponsoring Agency: Kriminalistik Verlag Gmbh
6900 Heidelburg 1, Germany
Format: Article
Language: German
Country: West Germany (Former)
Annotation: THE AUROS AUTOMATIC SPEAKER RECOGNITION SYSTEM WAS TESTED FOR RELIABILITY IN THIS EXPERIMENT BY THE WEST GERMAN MINISTRY OF RESEARCH AND TECHNOLOGY.
Abstract: AUTOMATIC VOICE RECOGNITION SYSTEMS THAT CAN BY CONNECTED TO THE PUBLIC TELEPHONE SYSTEM SERVE TWO FUNCTIONS: THEY VERIFY THE SPEAKER'S PROFESSED IDENTITY BY COMPARING THE TELEPHONE VOICE TO ALREADY STORED VOICE DATA OF THAT PERSON, AND THEY ATTEMPT TO IDENTIFY THE SPEAKER BY SCANNING THE DATA BASE FOR AN IDENTICAL VOICE. COMPUTERIZED VOICE RECOGNITION FACES TWO BASIC PROBLEMS: THE VARIATIONS OF THE VOICE PRINTS OF AN INDIVIDUAL SPEAKER AND THE HUGE AMOUNT OF DATA THAT MUST BE DECODED TO IDENTIFY A VOICE. THE AUROS (AUTOMATIC RECOGNITION OF SPEAKERS) SYSTEM TESTED IN THE PHILIPS LABORATORY IS ABLE TO IDENTIFY IN EACH CASE THE MOST EFFICIENT METHOD FOR IDENTIFICATION (E.G., THE FOURIER TRANSFORMATION, THE WALSH-HADAMARD TRANSFORMATION, OR INVERSE FILTERING). THE AUTOMATIC IDENTIFICATION PROCESS INCLUDES PRELIMINARY DATA TREATMENT, ANALYSIS OF THE SPEECH SEGMENT, CLASSIFICATION OF THE VOICE, AND CONCLUDING VERIFICATION. THE RELIABILITY RESEARCH ON AUROS INSPECTED 5,000 VOICE SAMPLES OF KNOWN SPEAKERS IN ACCORDANCE WITH THE MOST EFFICIENT IDENTIFICATION METHOD. THE SYSTEM WAS FORCED TO MAKE AN IDENTIFICATION, AND COMPUTER REJECTIONS WERE NOT ACCEPTED. FOR THE 2,500 CONTEXT DEPENDENT AND THE 2,500 CONTEXT INDEPENDENT SAMPLES, THE RECOGNITION RATE WAS AS HIGH AS 99.5 PERCENT. DURING THE VOICE VERIFICATION TEST, THE SYSTEM HAD TO STEER A MIDDLE COURSE TO AVOID TWO OPPOSING SOURCES OF ERROR: THE ERRONEOUS ACCEPTANCE OF A NONIDENTICAL VOICE AND THE ERRONEOUS REJECTION OF AN IDENTICAL VOICE. THE ERROR RATES IN BOTH DIRECTIONS WERE FOUND TO BE 1 PERCENT. THE PROBLEMS OF DELIBERATE VOICE DISTORTION AND OF DISTORTION THROUGH THE TELEPHONE SYSTEM ARE AS YET UNRESOLVED AND AS OF 1977 FORM THE SUBJECT OF A MAJOR RESEARCH PROJECT SPONSORED BY THE WEST GERMAN BUREAU OF CRIMINAL INVESTIGATION. THE ARTICLE CONTAINS NUMEROUS VOICE GRAPHS, MATHEMATICAL FORMULAS, AND A SHORT BIBLIOGRAPHY. --IN GERMAN. (SAJ)
Index Term(s): Computer aided operations; Crime laboratory equipment; Germany; Spectroscopy; Suspect identification; Testing and measurement; Voice identification
To cite this abstract, use the following link:
http://www.ncjrs.gov/App/publications/abstract.aspx?ID=66275

*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.