The problem of musical gesture continuation and a baseline system
Abstract
While musical gestures have been mapped to control synthesizers, tracked or recognized by machines to interact with sounds or musicians, one may wish to continue them automatically, in the same style as they have been initiated by a performer. A major challenge of musical gesture continuation lies in the ability to continue any gesture, without a priori knowledge. This gesture-based sound synthesis, as opposed to model-based synthesis, would open the way for performers to explore new means of expression and to define and play with even more sound modulations at the same time.
We define this new task and address it by a baseline continuation system. It has been designed in a non-parametric way to adapt to and mimic the initiated gesture, with no information on the kind of gesture. The analysis of the resulting gestures and the concern with evaluating the task raise a number of questions and open directions to develop works on musical gesture continuation.