Whenever that a musician plays a musical piece, the result is never a literal interpretation of the score. These performance deviations are intentional and constitute the essence of the musical communication. Deviations are usually thought of as conveying expressiveness. Two main purposes of musical expression are generally recognized: the clarification of the the musical structure and the transmission of affective content. The challenge of the computer music field when modeling expressiveness is to grasp the performers "touch", i.e., the musical knowledge applied when performing a score. One possible approach to tackle the problem is to try to make explicit this knowledge using musical experts. An alternative approach, much closer to the human observation-imitation process, is to directly work with the knowledge implicitly stored in musical recordings and let the system imitate these performances. This alternative approach, also called lazy learning, focus on locally approximating a complex target function when a new problem is presented to the system. Exploiting the notion of local similarity, the chapter presents how the Case-Based Reasoning methodology has been successfully applied to design different computer systems for musical expressive performance. © 2012 Springer-Verlag Berlin Heidelberg.
Arcos, J. L. (2012). Music and similarity based reasoning. Studies in Fuzziness and Soft Computing, 273, 467–478. https://doi.org/10.1007/978-3-642-24672-2_24