With the explosive growth of digital music data\nbeing stored and easily reachable on the cloud, as well as\nthe increased interest in affective and cognitive computing,\nidentifying composers based on their musical work is an\ninteresting challenge for machine learning and artificial intelligence\nto explore. Capturing style and recognizing music\ncomposers have always been perceived reserved for trained\nmusical ears. While there have been many researchers targeting\nmusic genre classification for improved recommendation\nsystems and listener experience, few works have addressed\nautomatic recognition of classical piano composers as proposed\nin this paper. This paper discusses the applicability of\nn-grams on MIDI music scores coupled with rhythmic features\nfor feature extraction specifically of multi-voice scores.\nIn addition, cortical algorithms (CA) are adapted to reduce\nthe large feature set obtained as well as to efficiently identify\ncomposers in a supervised manner. When used to classify\nunknown composers and capture different styles, our proposed\napproach achieved a recognition rate of 94.4% on\na home grown database of 1197 pieces with only 0.1% of\nthe 231,542 generated featuresââ?¬â?which motivates follow-on\nresearch. The retained most significant features, indeed, provided\ninteresting conclusions on capturing music style of\npiano composers.
Loading....