Automated analysis of postural and movement qualities of violin players

Learning to play a music instrument is a complex task, requiring continuous practice and the development of sophisticated motor control techniques. The traditional model of music learning is based on a master-apprentice relationship, leading often to a solitary learning process, in which the time spent with the teacher is usually limited to weekly lessons and a long period of self-study is needed. Moreover, a large amount of time passes from the teacher's feedback and the student's proprioceptive perception while studying, requiring a big effort in developing an efficient and healthy technique. In this paper, we present our recent developments concerning an assistive and adaptive technology to help violin students overcoming all these difficulties, and developing their technique and repertoire properly and safely. In particular, we focus on the multimodal corpus of violin performances which was collected for the purpose, and on the analysis of such data to compute postural and gestural features characterizing the performance under a biomechanical perspective and in terms of movement quality. Analysis is expected to provide students with feedback for reaching a physically accurate performance, maximizing efficiency and minimizing injuries.