Generative music is a form of music in which a piece of music creates itself from an initial set of musical elements and behaviours and rules defined by the composer and/or a system (natural or artificial). It is an “approach to music creation concerning itself with neither improvisation nor explicit composition, but rather with framing an indeterminate system from which music can emerge” (Priestley, 2014, p. 1). It is therefore not a musical genre or style on its own but rather a compositional practice where the composer is more concerned with creating or discovering a system or a process – physical or virtual – that will then generate the music autonomously of the composer, than with writing the composition from start to finish in the traditional sense. In generative music the role of the composer could be seen more as that of a gardener than an architect, to use Brian Eno’s analogy (Edge, November 10, 2011). Or to borrow another metaphor of his, “generative music is like trying to create a seed, as opposed to classical composition which is like trying to engineer a tree” (Toop, 2004, p. 182).
The term ‘Generative Music’ was originally used and popularised by Brian Eno with his 1996 release Generative Music 1. This was a floppy disk release for PC computers with certain kind of soundcards (Creative Labs AWE32 or SB32 soundcard or TDK MusicCard, to be precise), and it would generate endless variations of 12 compositions created by Eno and made with SSEYO Koan software (Intermorphic, n.d.). Inspired by cybernetics and systems theory, Eno had been using generative processes and systems in his music for records and installations throughout his career, starting with the 1975 release Discreet Music: these often consisted of analogue systems like tape loops of differing lengths or sets of CD players in a shuffle mode playing simultaneously, resulting in everchanging, indeterminate compositions. From Generative Music 1 onwards these systems have become increasingly digital and more elaborate, realised with dedicated software like those by Intermorphic (the developers of the aforementioned Koan) or through computer algorithms using the Objective-C programming language, eg. for his apps Bloom, Trope and Scape with Peter Chilvers (Digicult, n.d.).
Ideas similar to generative music, however, predate Generative Music 1 at least by two thousand years to ancient Greece: the Aeolian harp was “played” by the movement of wind over its strings, which initiated harmonic resonances to create the harp’s eerie sound (Hankins & Silverman, 1995). Later in the Middle Ages various algorithmic and mathematical methods were used to generate melodies and permutations of rhythmic and melodic patterns: the earliest known example of an algorithmic composition dates from the 11th century by Italian composer and music theorist Guido D’Arezzo, who ca. 1026 developed a method for mapping the vowels of a text to a set of pitches to generate melodies (Diaz-Jerez, 2000); and isorhythmic motets of the 14th and 15th centuries used the repetition of rhythmic and melodic patterns of differing lengths for the voice parts, resulting in numerous possible rhythm-melody combinations (Diaz-Jerez, 2000) – an idea which would resurface in the 20th century in the form of process music and tape loops. In the classical period the Musikalisches Würfelspiel (“musical dice game”) became a relatively popular system among composers to randomly generate music from precomposed measures using dice: the technique was pioneered by Johann Philipp Kirnberger in 1757, and one of its most well-known versions is Mozart’s manuscript K. 516f from 1787 (Diaz-Jerez, 2000). But while Mozart’s sequence of measures generated through chance still had to conform to the stylistic rules of the time, John Cage, using similar aleatoric procedures two centuries later, was free to break every stylistic convention of his time. Alongside Cage, the use of chance and randomness as a compositional technique became commonplace among the 20th century avant-garde composers, among them Charles Ives, Henry Cowell, Earle Brown, Cornelius Cardew and Pierre Boulez, as did the idea of music emerging from a process – the music being the process itself rather than a fixed score – like in the works of Elliot Carter, Alvin Lucier, Karlheinz Stockhausen, Terry Riley and Steve Reich. Similarly, the use of natural and scientific systems as means to generate music began to take hold: from Joseph Schillinger’s System of Musical Composition in 1920s and 30s5 to Edgard Varèse’s science-inspired organization of electronic sound, from the stochastic processes of Iannis Xenakis in the 1950s to the application of information theory to music by Lejaren Hiller, whose 1957 composition The Illiac Suite is allegedly the first composition created with a computer. This art-science approach to music by the 20th century composers could be seen as a revival of the idea of music proposed by the ancient Greeks (Diaz-Jerez, 2000).
‘Generative music’ is often used interchangeably with ‘algorithmic music’, given that a lot of contemporary generative music is being created with the help of computer-generated algorithms (Collins, 2008). And while there is a good argument for the methodological clarity of music related research to refer primarily to ‘algorithmic composition’ when speaking of automated compositional processes (Pearce, Meredith, & Wiggins, 2002), it should be emphasised for the sake of a more comprehensive definition of generative music and art that they also include non-computer, nonalgorithmic systems and works. According to Philip Galanter, “generative art refers to any art practice where the artist uses a system, such as a set of natural language rules, a computer program, a machine, or other procedural invention, which is set into motion with some degree of autonomy contributing to or resulting in a completed work of art” (2003, p. 4). A procedural invention can include, for example, “a chemical reaction, the use of living plants, condensation and crystallization processes, melting substances, or any other physical process that can take place autonomously” (Galanter, 2006, p. 1). Personally I prefer the term ‘generative music’ even when discussing computer-generated algorithmic music: it is the idea of a more universal concept of generative processes – biological, social and technological – that the term implies that appeals to me.
Collins, N. (2008). The Analysis of Generative Music Programs. Organised Sound, 13(3), 237-248. https://doi.org/10.1017/S1355771808000332
Diaz-Jerez, G. (2000). Algorithmic Music: Using mathematical models in music composition (Doctoral dissertation, Manhattan School of Music). Retrieved from https://www.coursehero.com/file/13859288/Gustavo-Diaz-Jerez-DMA-Thesis-MSM/
Digicult. (n.d.). Peter Chilvers: Visual And Tactical Music. Retrieved from http://digicult.it/digimag/issue-048/peter-chilvers-visual-and-tactical-music/
Edge. (November 10, 2011). Brian Eno: Composers as Gardeners. Retrieved from https://www.edge.org/conversation/brian_eno-composers-as-gardeners
Galanter, P. (2003). What is Generative Art? Complexity Theory as a Context for Art Theory.
Galanter, P. (2006). Generative Art and Rules-Based Art.
Hankins, T. L., & Silverman, R. J. (1995). Instruments and the imagination. Princeton, NJ: Princeton University Press.
Intermorphic. (n.d.). Generative Music. Retrieved from https://intermorphic.com/sseyo/koan/generativemusic1/
Pearce, M., Meredith, D., and Wiggins, G. (2002). Motivations and Methodologies for Automation of the Compositional Process. Musicae Scientiae, 6(2), 119–47.
Priestley, J. (2014). Poiesthetic play in generative music (Doctoral dissertation, Virginia Commonwealth University). Retrieved from https://scholarscompass.vcu.edu/cgi/viewcontent.cgi? article=4402&context=etd
Toop, D. (2004). Haunted weather: Music, silence and memory. London: Serpent’s Tail.