Human-robot interaction based on human emotions extracted from speech

Emotions represent a mental state of psychological arousal and are manifested by expression or display of distinctive somatic and autonomic responses. Emotion perception is still an open problem, especially in human-robot interaction. This paper presents a robot used for human-robot interaction, capable of classifying emotions. In this research sound features of human voice are used to classify two categories of human emotion evaluation: positive and negative. Based on their ranking, only the best sound features are considered in the emotion classifier built in the robot. A great linear relationship and similarity of the perception on emotion evaluation of the robotic system and the human evaluator is shown.