Catálogo de publicaciones - libros

The aim of this chapter is twofold: to provide a succinct introduction to evolutionary computing, outlining the main technical details, and to raise issues pertinent to musical applications of the methodology. Thus this chapter should furnish readers with the necessary background needed to understand the remaining chapters in this volume as well as open up a number of important themes relevant to this collection.

If the preceding chapter was an introduction to evolutionary computation (EC) for musicians, this chapter is intended as an introduction to music as a problem domain for EC researchers. Since we cannot hope to provide even a bare-bones treatise on music appreciation, much less music theory, we assume that the reader is at least somewhat familiar with music, if not as a producer, at least as a consumer. We will start by trying to define some musical terms to work with, including ‘music’ itself, which will lead us to a brief excursion into human-computer interaction as a metaphor for musical performance. We will then conduct an informal task analysis of music to define the tasks musicians perform and survey how EC has been applied to facilitate (or obfuscate, in some cases) the performance of those tasks. We will then summarize the various approaches that have been taken in representation, fitness and genetic operators.

Replicating musical instruments is a classic problem in computer music.Asystematic collection of instrument designs for each of the main synthesis methods has long been the El Dorado of the computer music community. Here is what James Moorer, the pioneering computer music researcher at Stanford University and later director of the audio project at Lucasfilm, had to say about it (Roads 1982):

… But what if the synthesizer just ‘grew’ programs? If you pressed a ‘randomize’ button which then set any of the thousand ‘black-box’ parameters to various values and gave you sixteen variations. You listen to each of those and then press on one or two of them—your favourite choices. Immediately, the machine generates 16 more variations based on the ‘parents’ you’ve selected. You choose again. And so on. The attraction of this idea is that one could navigate through very large design spaces without necessarily having any idea at all of how any of these things were being made. (Eno 1996)

It is commonly agreed in the context of Western tonal music that expression is conveyed by delicate deviations of the notated musical score, through shaping physical parameters of performance, such as timing, loudness, tempo and articulation. Expressive music performance research is aimed at establishing why, where and how these deviations take place in a piece of music. Interestingly, even though there are many commonalities in performance practices, these deviations can vary substantially from performance to performance, even when a performer plays the same piece of music more than once.

This chapter describes the author’s ongoing work with evolutionary computation in the composing of ‘art’ or ‘concert’ music. Over the course of many years, the author has written and rewritten a computer program called GenDash that employs evolutionary computation. GenDash has been used to help compose pieces ranging from works scored for solo human speaker to string quartets to orchestral works to pieces for instrumentalist and electronic computer music to operas.

Imagine you are walking down the street past a coffeehouse that features live jazz. From inside the coffeehouse you hear a jazz quartet begin to play a tune. As you pause outside to listen, it sounds like a tenor sax player backed up by a standard jazz trio of piano, bass and drums. You recognize the tune as John Coltrane’s Giant Steps as the tenor player plays the song’s original melody in the first chorus of the tune. Once this ‘head’ chorus is complete, everyone continues playing in the second chorus, but the tenor player plays a melody that is decidedly not the original melody of the song, switching from the half note rhythm of the original melody to a more active eighth-note-based rhythm. The piano, bass, and drums seem to be playing things that are similar to what they played on the first chorus, except that the bass player is playing a note on every beat instead of roughly every other beat, and the drummer is more active and assertive. This continues for four more improvized choruses, at which point the tenor player begins playing the original melody of the tune again. After this reprise of the tune’s head, there is a brief coda and the tune ends.

Cellular automata (CA) are tools for computational modelling widely used to model systems that change some feature with time. They are suitable for modelling dynamic systems in which space and time are discrete, and quantities take on a finite set of discrete values. CA are highly suitable for modelling music: music is fundamentally time-based and it can be thought of as a system in which a finite set of discrete values (e.g. musical notes, rhythms, etc.) evolve in space and time.

Music is a pattern of sounds in time. A swarm is a dynamic pattern of individuals in space. The structure of a musical composition is shaped in advance of the performance, but the organization of a swarm is emergent, without pre-planning. What use, therefore, might swarms have in music?

The beginning of Chapter 2 offered a sensible definition of music as temporally organized sound . In the broader sense of this definition, one could arguably state that music is not uniquely human. A number of other animals also seem to have music of some sort. Complex vocalizations can be found in many birds (Marler and Slabbekoorn 2004), as well as in mammals such as whales (Payne and McVay 1971) and bats (Behr and von Helversen 2004). In a chapter suggestively entitled ‘Zoomusicologie’ in the book Musique, Mythe, Nature ou Les Dauphins d’Arion , Mâche (1991) presents an interesting discussion on the formal sophistication of various birdcalls. Recently Holy and Guo (2005) demonstrated that the ultrasonic vocalizations that male mice produce when they encounter female mice or their pheromones have the characteristics of song. What is intriguing is that primates who are close related to humans are not as ‘musical’ as those mammals that are far more distantly related to us. This intriguing fact suggests that music might have evolved independently among various types of animals, at various degrees of sophistication. In this context, it would be perfectly plausible to suggest the notion that robots might also be able to evolve music.