HARMONIC, MELODIC, AND FUNCTIONAL AUTOMATIC ANALYSIS Placido´ R. Illescas, David Rizo, Jose´ M. Inesta˜ Dept. Lenguajes y Sistemas Informaticos´ Universidad de Alicante, E-03071 Alicante, Spain
[email protected] fdrizo,
[email protected] ABSTRACT Besides the interpretation, there are many applications of this automatic analysis to diverse areas of music: edu- This work is an effort towards the development of a cation, score reduction, pitch spelling, harmonic compar- system for the automation of traditional harmonic analy- ison of works [10], etc. sis of polyphonic scores in symbolic format. A number of stages have been designed in this procedure: melodic The automatic tonal analysis has been tackled under analysis of harmonic and non-harmonic tones, vertical har- different approaches and objectives. Some works use gram- monic analysis, tonality, and tonal functions. All these in- mars to solve the problem [3, 14], or an expert system [6]. formations are represented as a weighted directed acyclic There are probabilistic models like that in [11] and others graph. The best possible analysis is the path that maxi- based on preference rules or scoring techniques [9, 13]. mizes the sum of weights in the graph, obtained through The work in [4] tries to solve the problem using neural a dynamic programming algorithm. The feasibility of the networks. Maybe the best effort so far, from our point proposed approach has been tested on six J. S. Bach’s har- of view is that of Taube [12], who solves the problem by monized chorales. means of model matching. A more comprehensive review of these works can be found in [1].