Generating structured music using quality metrics based on Markov models

摘要: In this research, a first order Markov model is built from corpus of bagana music, atraditional lyre Ethiopia. Different ways in which low models can beused to build quality assessment metrics for an optimization algorithm are explained. Theseare then implemented variable neighbourhood search that generates baganamusic. The results examined and thorougly evaluated. Due the size many datasetsit often only possible get rich reliable statistics models, yet these donot handle structure very well their output repetitive. A method proposedthat allows enforcement repetition within thus handling long termcoherence with model.