Interactive Music and Synchronous Reactive Programming

TL;DR

This paper introduces Skini, a synchronous reactive programming environment for live interactive music performances, enabling composers to balance deterministic scores with audience-driven nondeterminism, fostering artistic experimentation.

Contribution

It presents a novel programming methodology using HipHop.js for composing interactive scores that are executable in live settings, bridging music composition and reactive programming.

Findings

Skini supports diverse musical styles including classical, techno, and jazz.

Scores programmed in HipHop.js can be executed live with audience interaction.

The system allows for artistic variation while maintaining musical coherence.

Abstract

This paper presents Skini, a programming methodology and an execution environment for interactive structured music. With this system, the composer programs his scores in the HipHop.js synchronous reactive language. They are then executed, or played, in live concerts, in interaction with the audience. The system aims at helping composers to find a good balance between the determinism of the compositions and the nondeterminism of the interactions with the public. Each execution of a Skini score yields to a different but aesthetically consistent interpretation. This work raises many questions in the musical fields. How to combine composition and interaction? How to control the musical style when the audience influences what is to play next? What are the possible connections with generative music? These are important questions for the Skini system but they are out of the scope of this paper that focuses exclusively on the computer science aspects of the system. From that perspective, the main questions are how to program the scores and in which language? General purpose languages are inappropriate because their elementary constructs (i.e., variables, functions, loops, etc.) do not match the constructions needed to express music and musical constraints. We show that synchronous programming languages are a much better fit because they rely on temporal constructs that can be directly used to represent musical scores and because their malleability enables composers to experiment easily with artistic variations of their initial scores. The paper mostly focuses on scores programming. It exposes the process a composer should follow from his very first musical intuitions up to the generation of a musical artifact. The paper presents some excerpts of the programming of a classical music composition that it then precisely relates to an actual recording. Examples of techno music and jazz are also presented, with audio artifact, to demonstrate the versatility of the system. Finally, brief presentations of past live concerts are presented as an evidence of viability of the system.