Forensic Voice and Tape Analysis
Methods of forensic voice and tape analysis first entered the limelight during the Watergate scandal in the early 1970s, and the basic methodology—if not the tools and precision with which the techniques are practiced—has changed little since. Much of this field is concerned with identification or elimination using voice-stress analysis, but controversy over techniques and their admissibility as evidence remains. This disagreement, even among specialists, came to the forefront as forensic scientists on both sides of the Atlantic studied tapes allegedly released by terrorist mastermind Osama bin Laden in the fall of 2002.
Early history. Spectrographic analysis and related techniques make it possible to match a suspect to an incriminating sample of his or her speech—a threatening phone call, for instance, or a taped admission of guilt. Voice and tape analysis can also be used to clear a suspect. In this field, a scientifically verified match between a suspect (or another individual) and a voice sample is known as an identification, while scientific proof, by means of voice analysis, that a suspect and a voice sample do not match is called an elimination.
The U.S. Federal Bureau of Investigation (FBI) used spectrographic or voice identification analysis as early as the 1950s, but the technique did not gain scientific acceptance until a 1962 study by Lawrence Kersta, a researcher working with a 1940s-model Bell Laboratory sound spectrograph. Kersta maintained that "voiceprints," a term he coined, provide a unique means of identifying individuals. He went on to establish a professional association, the International Association of Voice Identification, which in 1980, became part of the more general International Association for Identification.
The 1970s. The word "voiceprint" would later be discarded, due to the false association with fingerprinting, which is a much more exact science. Nevertheless, spectrographic techniques continued to gain respect in forensic and law enforcement circles, thanks in part to a 1972 study at Michigan State University. The study found an error rate of two percent for false identification (instances in which the examiner chose the wrong match, or found a match when none existed), and five percent for false elimination, in which the examiner failed to recognize that a match existed. In the years immediately following this study, spectrographic techniques came to widespread public attention during the examination of tapes made by President Richard M. Nixon in the White House.
In 1979, a National Research Council committee presented the FBI with the results of a study on spectrographic voice identification under forensic conditions, involving some 2,000 forensic comparisons made by FBI personnel. The researchers' findings confirmed the impression that, while it was not an exact science, voice analysis could be useful. According to the study, error rates varied as a function of the properties of the voice studied, the conditions and techniques used, and the examiners' skills and knowledge.
Voice analysis today. The period since the 1970s has seen considerable evolution in spectrographic analysis and the related methods and tools, which include evidence handling, critical listening, magnetic development, waveform analysis, spectrum analysis, tape enhancement, and speed correction. The core methodology, however, remains the same; from machine readings of stress and other patterns in a subject's voice, a graphic representation is made so as to illustrate patterns of frequency, intensity, pitch, and inflection. Analysts use a two-step process, first the aural or listening stage, then the visual stage, which involves looking over the spectrograms or readouts.
Spectrographic analysis remains controversial. It is permissible as evidence in 35 of 50 states, and has a status—both in the eyes of the law and of professionals—akin to that of polygraphy or lie detection; although not perfectly reliable, it can be a helpful tool for screening suspects. Controversy over spectrographic analysis came to the forefront in November, 2002, when the Arabic news station al Jazeera released a recording of an alleged telephone call from bin Laden.
Analysts working for the U.S. Central Intelligence and National Security agencies studied, and verified the authenticity of, the tape, in which the voice spoke of recent terrorist actions and promised to unleash more attacks. At the Institute for Perceptual Artificial Intelligence in Switzerland, however, researchers were not as certain. Using biometric software, they judged it a 55%–60% likelihood that the tape was not genuine.
█ FURTHER READING:
Gardner, Robert. Crime Lab 101: Experimenting with Crime Detection. New York: Walker, 1992.
Ross, David F., and J. Don Read. Adult Eyewitness Testimony: Current Trends and Developments. New York: Press Syndicate of the University of Cambridge, 1994.
Saferstein, Richard. Criminalistics: An Introduction to Forensic Science. NJ: Prentice Hall, 1998.
Romanko, J. R. "Truth Extraction." New York Times Magazine. (November 19, 2000): 54.
Sachs, Jessica Snyder. Graphing the Voice of Terror. Popular Science. < http://www.popsci.com/popsci/science/article/0,12543,426271,00.html > (April 13, 2003).