Music rhythm tagging: New applications for music information retrieval and musical intelligence

摘要

This paper presents new applications for musical information retrieval and musical intelligence. Music rhythm tagging focuses on the retrieval of symbolic music pieces of notes missing relative duration values. That is, marking up the notes with reasonable beat lengths based on the musical context. A multi-dimension vector markov model is presented. A relatively correct duration measure is presented and the mean accuracy is close to 0.63 in the experiment.