Voice stress analysis

Principle and origins

Although the vocal microtremor (changes in the frequency of modulation of the fundamental frequency of the voice) has been studied and/or validated by various scientists , its initial discovery was in 1970 by Olaf Lippold. Lippold discerned that the 8-12 Hz range physiological tremor associated with the contraction of human muscles indicated a connection between psychological stress and its effect on the human body. On September 5, 1972, three retired U.S. Army officers—Allan D. Bell Jr., Wilson H. Ford, and Charles R. McQuiston—filed a U.S. patent under their corporation name, Dektor Counterintelligence and Security, Inc. for their “physiological response analysis method and apparatus.” . Bell's expertise was in counterintelligence, Ford's was in electronics, and McQuiston's was in polygraphy. Ford had invented an electronic device that utilized the theory of Lippold, Halliday, and Redfearn in which he tape-recorded the human voice, slowed it down to one-third or one-fourth its normal rate, and fed it through several low pass filters which then fed the signal into an EKG strip chart recorder. The strip chart recorder then made chart tracings on heat-sensitive paper. Their Psychological Stress Evaluator (PSE) was the first commercially available VSA. According to Allan Bell Enterprises, "All lie-detection examinations or evaluations are predicated upon the fact that telling a significant lie will produce some degree of psychological stress. Psychological stress, in turn, causes a number of physiological changes.". Advances in computer technology introduced the Computer Voice Stress Analyzer (CVSA). The CVSA records the human voice using a microphone, typically clipped to a shirt collar, unlike the polygraph, which measures physiological responses requiring numerous sensors, including those designed for countermeasures. The CVSA can also be used to analyze recent or archival audio and video recordings.

Academic analysis

Despite the overall effectiveness of Voice Stress Analysis when used by police agencies over the last 40 years, as with other technologies that have enabled a paradigm shift in the field of deception detection/truth verification, it is not surprising that its use has been controversial. Discussions about the application of Voice Stress Analysis have focused on whether this technology can indeed reliably detect stress, and, if so, whether deception can be inferred from the latter. Since modern VSA technologies (e.g., CVSA) utilize algorithms that provide computer-generated objective outputs (i.e., stress indicated; no-stress indicated) linked to specific questions, the subjective input of the observer (i.e., inquisitor) is a less interfering factor; therefore, the potential inference problem of “Othello error” is attenuated. The possible issue of sampling or other forms of population bias might be problematic for study designs whereby respective researchers pick subjects casually. However, recent CVSA studies have specifically employed blinded techniques to help abrogate this type of bias. It has also been argued that—even if stress could reliably be measured from the voice—this would be highly similar to measuring stress with the polygraph, for example, and that all critiques centered on polygraph testing apply to VSA as well. However, due to the inherent disparities of the technologies (i.e., CVSA is easier to use; takes less time per exam; is less expensive; is noninvasive; and works with voice recordings as well as live interactions, since results from voice recordings are reproducible) this argument is not completely sound. Despite the latter, however, both technologies have independently been found to be effective, especially in the hands of well-trained examiners who understand the importance of using properly developed questions in their investigations. In fact, many investigators utilize both decision support technologies intermittently as tools to enable the resolution of cases.

Historically, most investigations of VSA have been funded by the US Government and were conducted either by polygraph researchers or researchers selected by various proponents of the US Federal Polygraph program. Several of these studies reported that VSA performed at chance level when detecting deception. For example, one study team tested VSA on the recordings of interrogation of 74 suspects. Eighteen of these suspects later confessed, making the deception the most likely ground truth. With 48% accurate classification, VSA performed at chance level. Other published research showed similar results. In 2003, the National Research Council concluded, “Overall, this research and the few controlled tests conducted over the past decade offer little or no scientific basis for the use of the computer voice stress analyzer or similar voice measurement instruments”.

Regardless of the latter, globally, a sufficient number of well-designed and published studies reveal that the use of VSA, especially when utilized in real-life settings that encompass high-stakes consequence and jeopardy, are highly reliable and accurate when used as a decision-support tool by well-trained examiners. Recently conducted studies identified that the detection of deception is possible with the use of voice stress analysis software loaded onto a laptop computer. Such a study was published December 7, 2013, in the open-access journal, International Journal of Electrical, Electronics and Computer Engineering, whereby scientists in India found that Voice Stress Analysis (VSA) technology identified emotional stress better than another popular deception-detection technology.

Similarly, in 2012, the peer-reviewed Ukrainian Journal, Criminalistics and Court Expertise, published a unique, retrospective field study authored by James Chapman (Professor Emeritus, Former Director of Forensic Crime Laboratory, State University of New York at Corning) and Marigo Stathis (neurocognitive scientist and research analyst). The retrospective study evaluated the use of the voice stress analysis technology for the detection of stress associated with possible deception and was based on 18-years of criminal cases where VSA was used as a truth verification tool when all other investigative courses of action had been exhausted. The authors demonstrated that, as a result of using a combinatorial approach of VSA and a standardized questioning process, “Of the 329 confession possibilities, 92.1% of the CVSA examinations produced a “Stress Indicated” result, and 89% of those resulted in validated confessions. Most notably, in 96.4% of interviews conducted, where the CVSA indicated stress, suspects made self-incriminating confessions.” There were no cases wherein a confession was obtained in the absence of stress (despite the interrogation of those suspects). In particular, the most considerable stress levels were detected during the investigation of murder, grand larceny, and sexual crimes. To disprove the tenet that the CVSA is a prop, Bernoulli distribution was used for one featured case of Chapman’s study involving 20 suspects.

Investigators who convened at the Proceedings of the 2005 Hawaii International Conference on System Sciences identified that VSA technology can identify stress better than chance with performance approaching that of current polygraph systems.

In a three-year study conducted in 2000 by the U.S. Air Force Research Laboratory in Rome, New York, it was determined that the voice stress units tested were able to recognize stress in the spoken voice. Additionally, these units performed equally whether the voice was a live test or a recorded one. The study also provided the caveat that caution should be taken when using voice stress analysis in that it should only be used as an investigative tool and not relied on for a case conclusion.

Notable examples of use

There are a few cited cases wherein the use of VSA technologies didn’t bode well for investigations and the people involved, contributing factors of which could have included improper examiner training and/or the use of inappropriate interrogation techniques. For example, the well-publicized murder of 12-year-old Stephanie Crowe took place in her bedroom inside her home at Escondido, California, sometime between the late hours of January 20, 1998 to early morning January 21, 1998. Her older brother Michael Crowe and two of his friends were initially charged with her murder, but were eventually declared to be factually innocent by a judge. Their confessions under police interrogation, which included the use of VSA, are regarded as classic examples of false confession. Also, there is the 2012 case in which George Zimmerman fatally shot Florida teenager Trayvon Martin. Zimmerman was given a voice stress analysis test by the police department of Sanford, Florida. He passed the test. A videotape of the test was publicly released in June 2012.

However, there are also instances where VSA technologies have been impressively successful. For instance, Mark Lunsford passed a CVSA examination conducted by the Citrus County Sheriff’s Office in Florida when questioned by the police about the disappearance of his nine-year-old daughter Jessica Lunsford. Jessica was kidnapped on February 24, 2005, and subsequently raped and murdered. Lunsford was also given a polygraph by investigators from the Florida Department of Law Enforcement (FDLE). Polygraph results showed Lunsford’s answers were “deceptive” on some of the questions. Lunsford was further subjected to additional polygraph examination by the FBI and the results were inconclusive. Lunsford was cleared when 46-year-old neighbor John Couey, a convicted sex offender, was arrested on March 14 after DNA evidence was found in his home.

In his book The Clapper Memo, author Bob McCarty documents the effective use of the CVSA as a tool used in interrogations and intelligence gathering by U.S. military intelligence and special operations forces in the Middle East. McCarty also brings to light the "war" on the CVSA by its opponents, which lead to a directive by former Under Secretary of Defense James Clapper for troops to stop using all non-polygraph-based technology for credibility assessment.

A recent ruling from a U.S. Federal court judge may require sex offenders to submit CVSA examinations throughout the post-release supervision process. Not long ago, a confession made following a voice stress examination was allowed to be used as evidence in a rape trial.