The Sound of Deception - What Makes a Speaker Credible?

The detection of deception in human speech is a difficult task but can be performed above chance level by human listeners even when only audio data is provided. Still, it is highly contested, which speech features could be used to help identify lies. In this study, we examined a set of phonetic and paralinguistic cues and their influence on the credibility of speech using an analysis-by-synthesis approach. 33 linguistically neutral utterances with different manipulated cues (unfilled pauses, phonation type, higher speech rate, tremolo and raised F0) were synthesized using articulatory synthesis. These utterances were presented to 50 subjects who were asked to choose the more credible utterance. From those choices, a credibility score was calculated for each cue. The results show a significant increase in credibility when a tremolo is inserted or the breathiness is increased, and a decrease in credibility when a pause is inserted or the F0 is raised. Other cues also had a significant, but less pronounced influence on the credibility while some only showed trends. In summary, the study showed that the credibility of a factually unverifiable utterance is in parts controlled by the presented paralinguistic cues.