A critical phonetic perspective on truth verification by advanced voice analysis technology

The ability of using reliable voice analysis technology to build advanced and non-invasive investigation tools for emotion detection and risk assessment would offer a number of important applications in areas of general public interest, like security and business. In the aftermath of the 2001 terror attacks against the World Trade Center, one such technology was patented and marketed in 2003.

Given the potential public importance of such a technological breakthrough, Anders Eriksson (University of Gothenburg) and I studied the working principles of the technology as described in the public patent but failed to find any evidence that the proposed method might work (Eriksson and Lacerda, 2007). Indeed, general acoustic, phonetic and signal processing principles strongly suggest that the technology carries out a rather crude and arbitrary voice analysis, given the physical characteristics of speech signals and the uncontrolled acoustic conditions in which they are registered (Lacerda, 2009).