How Does Voice-Stress Analysis Claim to Detect Lies?

Voice-stress analysis (VSA) is based on the idea that involuntary micro-tremors in the laryngeal muscles (around 8–12 Hz) are suppressed under psychological stress. VSA software analyses audio recordings for changes in these sub-audible frequency patterns, flagging responses as "stressed" or "unstressed." Proponents argue it can indicate deception without physical sensors. However, extensive peer-reviewed research — including studies by the U.S. Department of Justice and the National Research Council — has found VSA to be no more accurate than chance (about 50%) at detecting deception. Despite this, some law enforcement agencies and private companies continue to use it.