A Method for Comparative Analysis of Folk Music Based on Musical Feature Extraction and Neural Networks

A common problem in comparative musicology and ethnomusicology is that large collections of music are difficult to classify and visualize. Therefore, a tool which could be applied to either acoustic signals or symbolic representations would be useful. The choice of the features that are extracted from a collection of music and subsequently used by the tool should be psychologically relevant for the task. This study presents a simple data-mining tool for databases that use a symbolic representation of melodic information. The statistical distributions of melodic events are considered as a suitable features for several reasons. Firstly, the distributions are relatively straightforward to analyze computationally. Secondly, it has been shown that listeners are sensitive to pitch distributional information (Kessler et al. 1984; Oram & Cuddy, 1995; Krumhansl et al, 1999) and they can be used to predict similarity relationships between melodies (Eerola et al, 2001). It is also noteworthy that ethnomusicology has a long tradition in using statistical information to classify music (Freeman and Merriam, 1956; Lomax, 1968) and that there has been more recent attempts to classify musical styles according to their statistical features (Järvinen, Toiviainen, & Louhivuori, 1999).