Melody Extraction Method from MIDI Based on Polyphonic Clustering
MIDI is a digital interface for music,which can be called a score that a computer can understand,and the final part of a piece of music is the theme part,which contains the most important information in the piece.The purpose of theme extrac-tion is to classify different styles of music more accurately.In this paper,MIDI files will be selected and multi-track clustering algo-rithm will be used to extract the main melody of the music.Firstly,other accompaniment information in each track block is deleted to find the average pitch vector of each track block,then the track block is clustered and screened,and finally the main melody is extracted from the screened track block.By comparing with the standard theme,the algorithm can effectively extract the theme from the music,improve the accuracy,and provide a basis for the classification of music style.
国家科技期刊平台
NSTL
万方数据
