Accuracy of Voice Stress Analysis is Questionable
01/23/2009Can you outsmart a polygraph? The classic lie detection instrument that measures galvanic skin response, heart rate, and respiration has starred on various reality shows such as NBC's "Meet My Folks", in which parents screen potential dates for their daughter using the revered machine. In actuality, it is quite likely that the wrong suitors were eliminated from the show (and the daughter's life), given the fact that the accuracy of a polygraph is only a little more than chance, with most estimates around 60%. Polygraphs only measure physical responses. But what if they measured physiological responses, i.e., aspects of the liar - or truth-teller - that cannot be controlled? Voice stress analysis instruments do just that, but the supposed reliability of this newer method of deception detection may outshine the truth behind its accuracy.
People speak differently, with different vocal fold vibrations and differential articulatory styles, especially across regions, dialects, and languages. If voice stress analysis measures stress, a physiological response, and law enforcement officers in several states wholeheartedly approve of the technique for deception detection, then how, exactly, is stress linked to deception? One can imagine that being a suspect in a gruesome murder is a stressful situation, guilty or not. To be truly effective, voice stress analysis must be able to balance the stress of guilt with the stress of being in an unfavorable situation. However, because people experience stress as differently as they speak (perhaps more so), it does not seem prudent to apply the same technology to everyone equally and expect sufficient results to imprison someone.
A number of studies have been conducted on the accuracy of voice stress analysis instruments in detecting lies. A good number of the results garnered by groups not affiliated with the manufacturers of the technology have indicated that voice stress analysis is not much more effective than the polygraph machine. Critics of such studies defend their technology by concluding that the people conducting the studies were not properly trained in the use of the instrument, a process that took Lieutenant Bobby Martin and Major Bruce Temple, of the Halifax County Sheriff's Office in North Carolina one week at the National Institute for Truth Verification in Maryland.
Another worrisome topic related to voice stress analysis is its applicability. Casting aside the numerous doubts about its effectiveness in detecting deception, if the technology is as accurate as its proponents say it is, there remains the chance that it will be used in more than just a police station. What if employers screened phone calls from employees who called in sick? This could hurt the professional lives of the many well-meaning who falsely call in sick each year. What's worse, consider the case of a sick employee whose ailment distorts his voice, affecting the physiological aspects of his speech. His boss believes he is lying, and consequently fires him. That would be considered wrongful termination, which could end up costing the company thousands of dollars. Or, consider the suspicious spouse who thinks his significant other is cheating on him, when in fact, there is no foul play going on at all. A botched voice stress analysis could lead to the end of what could have been a long-lived and fruitful marriage. If voice stress analysis is to be adopted for widespread use, its applicability must be taken into ethical consideration.
Finally, those who choose to use voice stress analysis instruments for the detection of deception must exercise care. Widespread use of the technology is not a possibility, unless there is widespread training for its proper use. Despite the reservations expressed above, law enforcement officials have found the technology to be of use to them in determining the truthfulness of those subjected to the test, but it should not be the only factor that goes into an investigation. Just as man is fallible, machines can malfunction as well.
Perhaps all suspects involved in investigations should be considered truthful until proven deceitful.






