As a composer, I have always tried to combine the musical aspect with research aspects that I take into account in the framework of my research as a computer scientist. In this direction, I have developed some tools to be used when dealing with computer assisted music composition, that rely on algorithms I am using in such topics as Operation Research and Artificial Intelligence. Hence, when I have been asked to elaborate a sinthesis between Wagner and Mozart, I have promptly accepted, also in order to test the accuracy of such tools. The main idea is to look for some musical material by one of the two authors (i.e., a melody, a rythm, or even a simple chord) that minimises a given “distance” with respect to some material by the other author: when a coding has been defined for representing the material, the distance can be computed using well-known measures (Manhattan distance, euclidean distance etc.). Though this framework may seem rigorous, the composer action is of the utmost relevance: first, the coding has to be defined; then, the distance measure has to chosen and to be set; last, one have to decide what to do with the obtained results.
The three original compositions show us three different ways to handle the obtained results, keeping the other components fixed.
“San Marino” is based on looking a common feature between two pieces (melodies) having minimum distance, and using this common feature as input of a local search algorithm (Tabu Search) that explores a graph. The resulting music is short and neat, and the composer and pianist role is kept as low as possible.
Much different is “Sant’Agata”, in which the material previously found are used to define the harmonic base of the composition. Once this base is created, the same algorithm is used to select the musical material by browsing randomly generated graphs. This material is then freely used and transferred via “midi” to an automatic toy piano, belonging to the <M&M> Orchestra (Logos Foundation, Gent (B)).
“Tomé” instead, is an iprovisation based on clustering several melodic material by means of specific algorithms (K-Means). Each part of the composition corresponds to a given cluster. In order to decide the “mood” of each part I have considered some artworks of artists from the exibition (Bierbrauer, Enzweiler, Linschinger, Radoy, and Staudt). I wouldn’t say that there is a proper musical program behind the composition, but some choices have been “biased” after having seen the exibition itself.