Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.589202
Title: The acoustic and temporal characteristics of deceptive speech
Author: Kirchhuebel, Christin
Awarding Body: University of York
Current Institution: University of York
Date of Award: 2013
Availability of Full Text:
Access from EThOS:
Access from Institution:
Abstract:
Voice stress analysis based technologies, commonly known as Voice Stress Analysers (VSA), which are said to measure peoples’ veracity based on the speech signal, have come under a large amount of scientific scrutiny in recent years. Scientific reliability testing of these products has exclusively resulted in negative evaluations. While testing of these products is a necessary part of their evaluation, it is believed that a more fundamental step has been overlooked. Prior to examining the reliability of a test it should be ascertained whether the assumptions on which the test is based are valid. In other words, whether a relationship exists between deception, truth and speech, and if so, what the nature of this relationship is. Using two empirical studies, the present research explores the viability of using speech analysis as a means of differentiating between deceptive and non-deceptive speech. Experiment 1 takes the form of a laboratory-based study which employs a mock-theft paradigm in conjunction with a ‘security interview’ to elicit baseline/control and deceptive speech from a total of ten male native British English speakers. The data for experiment 2 is taken from audio recordings of the interrogation sessions that were part of a broader research study investigating a number of human deception responses across biological, physiological, psychological and behavioural dimensions. The recordings provided control, truthful and deceptive speech from a total of 37 speakers. The speech samples were analysed on a range of acoustic and temporal parameters. For the majority of acoustic parameters no significant differences or trends can be discerned that would allow for a reliable differentiation between deceptive and non-deceptive speech. The changes that do occur are very minute and not consistent across speakers. The results of the temporal measurements seem more promising with respect to highlighting differences between deceptive and non-deceptive samples. However, it transpired that contextual factors such as type of interview and nature of questions asked have conflicting effects on the temporal aspects of liars’ speech. Moreover, context appears to affect the speech of truth-tellers in similar ways to that of liars. Deceptive behaviour is individualised, very multifaceted and far from being clear cut. The successful development of a method that would reliably and consistently differentiate between truths and lies, based on speech analysis, is highly questionable.
Supervisor: Howard, David M. Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.589202  DOI: Not available
Share: