THE POTENTIAL OF A MUSIC-INSPIRED ALGORITHM FOR MUSIC DOCUMENT GROUPING

Abstract

Nowadays we can easily obtain a lot of music sound files or documents on the Internet. In this flood of music information, we want to sort them out in order to utilize them efficiently. This paper proposes a new algorithm, named Harmony Search (HS), which is inspired by the improvisation performance of music players. The HS algorithm numerically simulates the musical process whereby musicians improvise new harmonies from their experience or from complete randomness and improve them through experience. By doing this simulation, the HS algorithm could be successfully applied to diverse optimization problems such as Sudoku puzzles, music composition, and engineering design. This paper examines the possibility of applying HS to classifying music sound files or music-related documents, and suggests HS can be a useful module for efficient searching.