Markov Constraints Revisited: Commentary on “Markov Constraints: Steerable Generation of Markov Sequences”

By François Pachet and Pierre Roy

François Pachet and Pierre Roy. “Markov Constraints Revisited: Commentary on “Markov Constraints: Steerable Generation of Markov Sequences””. Constraints, 31, 140–147, 2026.

About the paper

A melody can sound like the examples a model learned from and still fail to end where the musician wants. This retrospective traces how the 2011 Markov Constraints paper addressed that gap by combining a model of musical style with rules for the whole sequence.

Abstract in brief

The commentary follows the original idea through applications to meter, lyrics, the avoidance of copied passages, exact sampling, and musical form. It also connects the work to Flow Machines and explains why explicit constraints remain useful for controlling neural generators.

Related research

Start with the original 2011 paper, or explore the Markov Constraints project and its browser playground. The Flow Machines project shows how the research was applied to music composition.