A Comparison of Language Modeling and Probabilistic Text Information Retrieval Approaches to Monophonic Music Retrieval

With interest in music information retrieval increasing the need for retrieval systems unique to music is also growing. Despite its unique properties music shares many similarities with text. The goal of this paper is to explore some of the capabilities and limitations of current text information retrieval systems as applied to the task of music retrieval. Monophonic music is converted into text and retrieval experiments are run using two different text information retrieval systems in various configurations. Finally, we will discuss whether the techniques applied here are generalizable to the larger problem of polyphonic music retrieval.