An Approach to Authenticity Speech Validation Through Facial Recognition and Artificial Intelligence Techniques

Hugo Faria, Manuel Rodrigues, Paulo Novais

Abstract: Since all times, humans tend to adapt their speech, in terms of authenticity, according to the moments specific needs, making some statements or claims about something that do not correspond to reality, in short, lying about some matter. Identifying such moments has always been a challenging task, not at all times successful, and requiring external artefacts, with scarce availability and inducing stressful situations. With recent advances in hardware technology, making enormous computational power available in our hands through smartphones and fast network technologies, and with Artificial Intelligence evolution, namely Deep Learning, it seems possible to achieve the goal of validating speech authenticity, with smartphones, using facial recognition to detect signs of untruthful speech. This paper presents a framework to achieve this goal.