NCJ Number
44206
Date Published
1976
Length
14 pages
Annotation
THE SENSITIVITY OF THE VOICEPRINT METHOD OF SPEAKER IDENTIFICATION TO INTENTIONAL DISGUISING OF SPEECH IS ASSESSED IN A STUDY USING 25 MALE SPEAKERS AND 6 EXPERT VOICEPRINT INTERPRETERS.
Abstract
THE SPEAKERS AND INTERPRETERS WERE FACULTY MEMBERS AND GRADUATE STUDENTS IN THE FIELD OF PHONETICS. THE SPEAKERS RECORDED A STANDARD PASSAGE TWICE, ONCE USING NORMAL VOICES AND AGAIN USING DISGUISED VOICES. THE INTERPRETERS ATTEMPTED TO MATCH THE DISGUISED-VOICE SPECTROGRAMS WITH THE NORMAL-VOICE SPECTROGRAMS. WITH FEW EXCEPTIONS, THE INTERPRETERS WERE UNABLE TO MATCH THE TWO SETS OF SPECTROGRAMS CORRECTLY. OFTEN A GIVEN SPEAKER'S DISGUISED-VOICE SPECTROGRAM RESEMBLED ANOTHER SPEAKER'S SPECTROGRAM MORE CLOSELY THAN THE GIVEN SPEAKER'S OWN NORMAL-VOICE SPECTROGRAM. THE FINDINGS ARE TAKEN AS FURTHER EVIDENCE OF THE SHORTCOMINGS OF THE VOICEPRINT AS A SPEAKER IDENTIFICATION METHOD. IT IS SUGGESTED THAT THE VOICEPRINT TECHNIQUE NOT BE USED IN INVESTIGATIVE OR FORENSIC SITUATIONS UNTIL MORE IS KNOWN ABOUT ITS VALIDITY. RESEARCH DIRECTIONS IN SPEAKER IDENTIFICATION ARE SUGGESTED.