Technology Enhanced Learning of Expressive Music Performance

Learning to play music is mostly based on the master-apprentice model in which modern technologies are rarely employed and students’ interaction and socialisation is often restricted to short and punctual contact with the teacher. This often makes musical learning a lonely experience, resulting in high abandonment rates. In the context of TELMI, an international project, which aims to addresses these issues by providing new multimodal interaction paradigms for music learning and to develop assistive, self-learning, real-time feedback, complementary to traditional teaching, this paper focuses on the computational modelling of expressive music performance as a tool for music learning. We record a professional violinist and apply machine learning techniques to induce an expressive model the recordings. We use this model to generate feedback on expressive aspects of arbitrary pieces to violin students.