Skip to main content icon/video/no-internet

Vocal stress analysis refers to computer-aided technology and techniques for analyzing nonverbal speech in order to aid deception detection. Modern vocal stress analysis is grounded in paralinguistics, digital signal processing, and machine learning. Vocal stress analysis is a term used most commonly in reference to software used by law enforcement as an alternative or complement to traditional polygraph tools for investigations.

To assess veracity using vocal stress analysis software, a speaker's voice is first recorded using a microphone and converted to a digital format on a computer. The vocal recording is then signal processed to extract acoustic features, such as pitch or intensity representations (for example, mean, variance) of each utterance. The utterances generally correspond with a fixed protocol comparable to the Comparison Question Test used in a polygraph. Based on the testing protocol the overall interview is scored and deviances or outliers are identified to assess veracity. This example is a simplification, but it provides the basic principles behind vocal stress analysis—record the voice digitally, extract vocal features, and submit these features to a statistical/computational model to assign a probability of deception.

Vocal stress is a term used almost ubiquitously in academic and professional settings when discussing software for assisting deception detection. However, vocal stress is actually a misnomer, as modern techniques for analyzing the voice analyze more than stress. Modern techniques and technologies for analyzing speech attempt to identify vocal indicators of stress, emotion, cognitions, and behavioral control. For example, paralinguistic indicators of cognitive effort are speech hesitations (that is, response latency) or dysfluencies (that is, the use of the sounds “um” or “er” and pauses).

Vocal analysis confined to stress has intuitive appeal. When a nervous or stressed speaker talks, there is a noticeable increase in the pitch or tremble of their voice. When stressed, individuals' muscles often tighten and clench in response. When the muscles about the vocal folds, which vibrate to affect the air and produce sound, become tense, they vibrate at a higher frequency. The frequency of the vibration or fundamental frequency is what people hear as vocal pitch. Because of the observable variation in speech in response to stress, and the belief that liars faced with consequences are more stressed, the initial focus of vocal analysis was on stress.

Microtremors and Vocal Analysis

The term vocal stress analysis was coined during the initial wave of software marketed to law enforcement. Despite the richness of features present in the voice, this initial software focused on a very small frequency band of 8 to 12 hertz. This is because the human body exhibits periodic contractions of the muscles known as microtremors on this narrow and low frequency range. These early vocal analysis systems attempted, unsuccessfully, to measure this frequency produced by the larynx muscles.

Early vocal analysis systems assumed that a reduction in the power of the microtremor frequency implies deception because it is caused by a stress-induced drop in blood pressure. The microtremors do occur at the low frequency range; however, existing recording technologies may not have the sensitivity required to accurately measure and subsequently calculate this low frequency. Additionally, even if microtremors can be measured via the voice, the relationship between lower blood pressure and deception is tenuous.

...

  • Loading...
locked icon

Sign in to access this content

Get a 30 day FREE TRIAL

  • Watch videos from a variety of sources bringing classroom topics to life
  • Read modern, diverse business cases
  • Explore hundreds of books and reference titles

Sage Recommends

We found other relevant content for you on other Sage platforms.

Loading