An approach to predicting bowing control parameter contours in violin performance

Abstract

We present a machine learning approach to modeling bowing control parameter/ncontours in violin performance. Using accurate sensing techniques/nwe obtain relevant timbre-related bowing control parameters such as bow/ntransversal velocity, bow pressing force, and bow-bridge distance of each/nperformed note. Each performed note is represented by a curve parameter/nvector and a number of note classes are defined. The principal components/nof the data represented by the set of curve parameter vectors are obtained/nfor each class. Once curve parameter vectors are expressed in the new space/ndefined by the principal components, we train a model based on inductive/nlogic programming, able to predict curve parameter vectors used for rendering/nbowing controls. We evaluate the prediction results and show the potential/nof the model by predicting bowing control parameter contours from an/nannotated input score.