A Live-learning Punitive Interface for Improvisational Performance Dynamics

Kevin Blackistone; Martin Kaltenbrunner

A Live-learning Punitive Interface for Improvisational Performance Dynamics
Image credit: Kevin Blackistone; Martin Kaltenbrunner

Abstract:

There have been many permutations in methods to train musicians in the performance technique and the technical skills required for an instrument. Among these, many modern computational approaches have been developed which incorporate some level of feedback from the student. These often assess mastery based on a response, dynamic or learned, to the practiced result. This research presents an interface expanding these concepts, which guides the performer to novel improvisation through punitive techniques. As opposed to models using prior composition or predetermined reference points, it live-learns in situ to assess predictability without pretraining. It provides a framework and interface which seeks to be implementable for a variety instruments and styles analyses of sound and pose.