Audio-Visual Lie Detection

Description Finely-grained behavioural analysis, based on audio and visual signals and machine learning has the potential to reveal differences between people telling the truth and people who are lying.
Task [THESIS ONLY] Utilising a unique dataset of individuals telling a truthful narrative and an untruthful analysis speech and visual cues will be explored for their usefulness in providing objective measurements for lie detection.
Utilises Python, openSMILE, openFACE
Requirements Reasonable knowledge of signal processing and machine learning
Languages English
Supervisor Dr. Nicholas Cummins (