A fear of public speaking can have a significant impact on an individual’s success in enterprise and education. This thesis presents a new, multimodal, digital system which enables users to practise their public speaking skills and gives them visual feedback in real time. It has been developed within the paradigm of Positive Computing, an interdisciplinary paradigm for human-computer interaction. Positive Computing has a number of themes including competence, self-awareness, stress-reduction and autonomy. The term ‘multimodal’ refers to the fact that the system detects multiple speaking modes in the speaker such as their gestures, voice and eye contact. The user can select if they want to receive feedback on all speaking modes or a subset of them. The system consists of a Microsoft Kinect 1 connected to a laptop. The Microsoft Kinect, a 3D depth camera, is used to sense the user’s body movements, facial expressions and voice. The user stands in front of the system and speaks. Users can choose to see themselves represented on the screen as either an avatar or live video. Real-time feedback is superimposed on their chosen representation in proximity to the area it relates to. The purpose of the feedback is to make the user aware of their speaking behaviour. The feedback is non-directive as the user can choose if they want to modify their behaviour. This system gives users the potential to develop skill and confidence before speaking in front of a live human audience. Users have reported that they enjoy using the system and can see its benefits. Results and descriptions of all user testing are reported in detail in this thesis.