Fake Emotion Detection Using Affective Cues and Speech Emotion Recognition for Improved Human Computer Interaction

Differentiating between real and fake emotions is an emerging research topic that is becoming increasingly important, especially when considering how it affects human-to-human interaction, as well as human computer interaction (HCI). This can aid humans in detecting fake emotions, consequently enhancing the ability to differentiate those who are genuine from those who are not in their social circles and everyday interactions. It can also help in improving HCI of applications. This study presents the development and evaluation of two deep learning models aimed at detecting fake emotions. The first model focuses on analyzing facial expressions, while the second model relies on speech signals. The first model utilizes a combination of Convolutional Neural Network and Long Short-Term Memory to analyze facial expressions. Experimental results demonstrate that the proposed model achieves an accuracy of 70%. The second model explores a novel approach by employing the same CNN-LSTM architecture but with different dimensions and hyperparameters. This model focuses on analyzing emotional speech to detect fake emotions and it achieved an accuracy of 96.93%.