Employing a Voice-Based Emotion-Recognition Function in a Social Chatbot to Foster Social and Emotional Learning Among Preschoolers

Abstract

This study will introduce a social chatbot to a child learning site to cultivate children’s understanding of emotions and strengthen their emotional expression skills using conversation scenarios between a child and social chatbot under the guidance of a kindergarten teacher to achieve emotional ease and adjustment effectiveness. The use of AI-aided emotional measurement is a completely new direction for academic research regarding the speech emotion and behavioral analysis of preschoolers. The benefits of this study are as follows. (a) The establishment of children’s emotional speech database and analysis module. Currently, EMO-DB and Beyond Verbal contain samples of speech emotions, but the analysis of children’s emotional speech has not been reported. The speech and conversation data collected from preschoolers in this study might be used to construct speech/emotion characteristic database, an emotional semantic identification database, an image/emotion analysis database, and an interactive script database designed for preschoolers to assist children’s emotional development, social learning, and provide references. (b) The development of a voice chatbot with EQ. The setting of this study can be used regarding a voice chatbot with EQ and applied in real-world learning sites, enabling the analysis of real-time emotions based on the emotional speeches and words in dialog with a social chatbot. Moreover, it can pro-vide appropriate conversations based on the different emotional states of the user. The social chatbot developed in this study might help nurture children’s EQ and cognitive abilities, as well as improve their emotional adjustment, management, and social support skills.