MIDI DÖNÜŞTÜRÜCÜ YAZILIMLARIN BAŞARI KARŞILAŞTIRMASI VE MATLAB’DA MÜZİK ANALİZİ

Abstract

Computational music analysis is a specifically fields of study where music and technology entwined together and completing each other. The purpose of analysis, besides its social scientific influences, is to process some data that cannot be detected with human possibilities. MatLab, that becoming a comprehensive laboratory for all other engineering production and solutions and that converting music data into computable values for all purposes, has become one of the most effective software for music analysis. Music analysis with Matlab can be applied by users over raw audio files, generally. But more commonly used method is to use symbolic data that represents to music. One of the strongest symbolic representations are MIDI. Matlab uses with a variety of toolboxes for music analysis such as MIDI ToolBox (MTB) that developed by Finnish scientists. Processing some data with hundreds of analysis functions, the MTB uses music files that have been completely converted to MIDI. At this point it is important to convert a music file to MIDI. In this study, some software that convert audio to MIDI data has been tested. Monophonic and polyphonic form of piano music (sonata) in classical period have been used as a source of converting software, and the obtained results are analyzed by f-measure method. The results show that the most effective way to create a MIDI file is to play or write directly. Even WIDI and Melodyn, considered the most successful software in the test, may cause incorrect or incomplete analyzes with these results.