14.5 Musical examples

Lejaren Hiller and Leonard Isaacson - Illiac Suite (1957)

The Illiac Suite for string quartet is one of the first pieces of music composed by a computer, the ILLIAC I at the University of Illinois. Each of its four movements, or “experiments”, explores a different algorithmic technique [HI59]. The fourth experiment composes with Markov chains, whose transition probabilities Hiller and Isaacson designed by hand, rather than fitting them to data. For example, one of their chains favors small steps between consecutive notes, while another favors consonant intervals.

Philippe Manoury - Pluton (1988)

In Pluton, for piano and live electronics, the computer listens to the pianist, and builds Markov chains from the notes they play, in real time. The chains then generate new music in the style of what the pianist just played, which the computer plays back by transposing recordings of the piano, captured on the fly. Pluton was one of the first pieces made with Max, the visual programming language for computer music that Miller Puckette developed at IRCAM.

David Cope - Mazurka, Op. 1, No. 2 (after Chopin) (1997)

David Cope’s Experiments in Musical Intelligence (EMI) composed new pieces in the styles of composers from Bach to Joplin, by recombining fragments of their works that respect the patterns characteristic of each composer. This mazurka, a Polish dance form, imitates Chopin, who wrote dozens of them. In this recording, Mary Jane Cope performs it on the piano. Some listeners have mistaken EMI’s imitations for the real thing.

Autechre - Untitled (2022)

Markov chains with hand-designed probabilities are still a tool of modern electronic music. For example, the duo Autechre is known for generative rhythms programmed in Max. In this untitled demo from 2022, a sparse transition table mostly follows a fixed rhythmic pattern, but encodes a few jumps that occasionally send the rhythm somewhere unexpected.