Music Melody Prediction using Hypernetworks)

Human can follow melody which he has not listened to before. Especially human predict next melody better if they are familiar with the genre of the song. We assume that melody is time series data. Recently hypernetworks have been researched for classification and time series problems. In this paper, we use evolutionary hypernetwork models (EHNs) for predicting difference of pitch. Evolutionary hypernetwork model can represent the corpus of melody and predict unseen melody partly.