Stand-alone application of user-specific emotion recognition model to improve real-time online voice communication using expressive avatar

One of the objectives of this research is to explore and investigate on how to improve online voice communication. We use our previously developed user-specific emotion recognition model to recognize user's emotion during communication and then to express it using an avatar to show to the partner. Another objective is to investigate the performance of our model in real-time environment using a stand-alone application developed by a party who is not proficient with emotion recognition. The results were: the said party can develop a stand-alone application despite not having much knowledge on emotion recognition using our model, and that real-time recognition can be achieved. The questionnaire results from the subjects also suggest that using expressive avatar improves communication, and that user-specific approach gives better performance compared to general approach.