How Did the Department of Defense Verify the Theory Behind Voice Stress Analysis?
In 1995, a new type of truth verification technology was making its mark in the law enforcement community: the Computer Voice Stress Analyzer (CVSA®). First introduced by the National Institute of Truth Verification (NITV) in 1988, the CVSA supplanted the Psychological Stress Evaluator (PSE), an earlier form of voice stress analysis. The PSE was introduced by Dektor Counterintelligence and Security, Inc. in 1972, but it achieved limited commercial success due to criticisms from the polygraph community, and the company went out of business in the 1990s.
With the PSE now obsolete and the CVSA becoming increasingly popular, the U.S. Department of Defense wanted to validate that the CVSA functioned in accordance with the theoretical claims of the manufacturers, from an electronics perspective. To answer this question, the Department of Defense assigned researcher Victor L. Cestaro to conduct a technical experiment analyzing the CVSA in a lab environment.
The Experiment: Testing the Theory Behind CVSA
Cestaro’s experiment was designed to evaluate and confirm the electronic theory of operation underlying the CVSA. According to the manufacturers, the CVSA was an effective means for detecting deception because its output (a visual chart representing the sound signal input) visibly changed in response to sounds with different fundamental frequencies. It is important to note that it researchers at University College London had already established that physiological tremor was associated with fundamental frequency, so Cestaro took this fact as a given and did not conduct further research into the physiological underpinnings of the CVSA.
According to established information regarding the CVSA, there were generally two types of traces that could appear on a CVSA chart. When the subject was not under stress, the tracings of the signal would appear “peaked” or “cyclic” on the page—picture an EKG chart featuring spikes and dips along a generally straight line, with a somewhat-defined upper and lower boundary. In contrast, the voice of a stressed subject would produce tracings with constant or near-constant amplitudes. On the page, the tracings would be confined to a much shorter range, so the trace would look more like a straight line.
To find out whether the CVSA really produced these unique patterns for sounds with different fundamental frequencies, Cestaro conducted two simulations using multiple lab function generators. These complex instruments can be used to simulate sound waves with different fundamental frequencies by directly inputting a sine wave signal into the CVSA microphone. For the unstressed voices, Cestaro used the generator to produce a signal with a frequency of about 10 Hz. For the stressed voices, Cestaro used the function generator to create unmodulated signals with a constant amplitude, at frequencies between 200 and 500 Hz.
At the different fundamental frequencies, Cestaro evaluated the output of the CVSA. Just as NITV had indicated, Cestaro observed that small changes in the frequency of the input signal directly impacted what the test evaluator would see on the page. Therefore, he was able to confirm the validity of the CVSA’s electronic output.
Cestaro’s report conclusively stated: “The lab simulations established that the CVSA performs electrically according to the manufacturer’s theory of operation and, even in the absence of jeopardy, which is a basic requirement in detection of deception.”
The main value of the 1995 Department of Defense study is that it validated the theoretical underpinnings of the CVSA from an electronics perspective. Researchers have shown law enforcement professionals rigorous scientific evidence linking the electrical principles of the technology and the real-world performance of the CVSA.
For law enforcement professionals, that is just one more confirmation of the technology’s usefulness in scenarios like criminal investigations, terrorist interrogations, and employment screenings, where the stakes are high and deception truly represents a risk.