Generating Music With An Optimization Algorithm Using A Markov Based Objective Function

In this paper, a machine learning approach is combined with an optimization algorithm to generate structured music for the bagana, an Ethiopian lyre. A low order Markov model that captures the style characteristics was built from a corpus of bagana music. Different ways are then explored in which a Markov model can be used to construct quality metrics that represent how well a fragment fits the chosen style. These metrics are implemented in the objective function of a variable neighbourhood search algorithm that generates bagana music.