Lie Detection using facial and voice recognition

Abstract

Our system will use multi-modal data from face and voice along with machine learning techniques to determine, to some degree of accuracy, whether someone has committed an act of deception. Our system will operate in two phases, the control phase and the critical phase. During the control phase, we will ask the subject baseline questions (questions that will produce answers that are true) and establish the normal behavior of the subject. We’ll train our system with the data obtained in the phase. During the critical phase, we’ll ask the subject questions that are likely to elicit a deceptive response and determine (in comparison to the normal behavior of the subject) whether a lie is being told.