Classifying Polyphonic Melodies by Estimate of Music Process
Yukiteru Yoshihara, Takao Miura, Isamu Shioya · 2007
In this investigation we propose a new and novel approach for classifying polyphonic melodies. Our main idea comes from Probability Stochastic Processes using Hidden Markov models where the characteristic features of polyphonic melodies are extracted from each bar. The similarity among harmonies can be considered by means of the features. We show the effectiveness and the usefulness of the approach by some experimental results.