Abstract
This work aims to improve the quality of visual speech synthesis by modelling its emotional characteristics. The emotion specific speech content is analysed based on the 3D video dataset of expressive speech. Preliminary results indicate a promising relation between the chosen features of visual speech and emotional content.