Modeling music as markov chains: Composer identification, https://www-ccrma.stanford.edu/~jacobliu/254report.pdf

In this research, it is proposed that music can be thought of as ran-dom processes, and music style recognition can be thought of as a system identification problem. Then, a general framework for modeling music us-ing Markov chains is described, and based on this framework, a two-way composer identification scheme is demonstrated. The scheme utilizes the Kullback-Leibler distance as the metric between distribution functions, and it is shown that under the condition when the marginals are identi-cal, the scheme gives maximum likelihood identification. Experiments of composer identification are conducted on all the string quartets written by Mozart and Haydn, and the results are documented and discussed.