DoDPI Banner for Research Papers

Abstract

CESTARO, V. L. and DOLLINS, A. B. An analysis of voice responses for the detection of deception. June 1994, Report No. DoDPI94-R- 0001. Department of Defense Polygraph Institute, Ft. McClellan, AL 36205.

This study was designed to examine the feasibility of using audio pitch analysis and spectrum decomposition techniques to aid in the detection of deception following a numbers test. Audio recordings were made of 44 male subjects' responses during a peak-of-tension (POT) test. A Lafayette field polygraph was used to collect respiration, cardiovascular, and electrodermal responses for manual evaluation. Half of the examinees were programmed "deceptive" and half were programmed "truthful". The audio recordings of the subjects' responses were analyzed off- line using pitch and spectral analysis software to examine differences between truthful and deceptive "no" responses. Useable voice recordings were obtained from 28 of the original 44 subjects. No significant differences were found between the two groups on individual measures of pitch variation, response duration, or mean response energy. A significant concurrence rate (p <.01) was seen between decisions made by pitch/energy analysis and the examiner's decisions based on analysis of the test data. Significant differences were found between the number of correct decisions made by the examiner (79%) and by pitch/energy analysis (37%). However, no significant differences were seen between the number of false positive decisions made by the examiner and by pitch/energy analysis (35% versus 29%).

Key-words: voice analysis, voice pitch, voice stress, detection of deception, fundamental frequency, spectrum analysis, vagal, parasympathetic, polygraph

Director's Foreword

This was the first study, in what will be a series of studies, to explore a variety of voice stress analysis methods and systems. The purpose of this line of research is to determine (1) if the data collected by any particular voice stress recording device, evaluated by a variety of visual and/or algorithm systems, can discriminate between individuals responding truthfully and those responding deceptively; (2) if detection of deception tests, using voice stress data, can supplement or supplant traditional approaches to psychophysiological detection of deception (PDD) tests; and (3) validity and reliability data for voice stress tests. Although the results of this study did not resolve or answer any of the above issues, further research is nevertheless warranted.

In recent years, there has been a revival of voice stress applications in PDD testing. In particular, the Computer Voice Stress Analyzer (CVSA) promoters have been actively advertising, selling and training examiners across the country. Agencies that have purchased the equipment; and, many that are contemplating the purchase of this equipment, are vitally interested in obtaining validity and reliability data that will support or reject the claims of CVSA promoters. Subsequent research projects will center on the CVSA and the test procedures developed for that instrument.

William J. Yankee, Ph.D.

Director