Automatic music teaching: A logic-probabilistic etude generator

Electronic publication.We explore the potential of automatic music generation as a tool in music teaching, to create musical études aimed at improving sight-reading or technique. The logic-probabilistic music generation system APOPCALEAPS can be parameterized to generate études in a particular style, but does not take into account the crucial didactic notion of difficulty. We describe a modified version of APOPCALEAPS which generates progressively more difficult pieces. As a case study, we restrict ourselves to sight-reading ́etudes in the blues style, for electric guitarists. Experiments show that our approach is very promising and that it correctly captures the notion of difficulty.status: publishe