Music for piano
and computer

A first exclusive preview of Duet was presented by KEF and Parameta in Hong Kong on November 28th and 29th, 2023 with the artist himself as the performing musician.

The accompanying NFT series was released on fxhash following the performance. It presents still images from the same parameter space as used for the preview performances.

About the piece

Duet is part of Andreas Rau’s ongoing exploration of human-machine collaboration. In a series of audiovisual generative performances starting in 2024, this body of work will be brought to life together with different musicians for audiences in locations around the world. Accompanying NFT series will immortalize the performances on the blockchain.

During the live performances, human musicians take control over the musical narrative: they decide when to start and end the performance, which notes to play, and which to leave out. The scores, however, are generated by the computer. In this way, Duet can be read as a metaphor for the interdependence of human and machine.

Human-machine collaboration

My interest in human-machine collaboration is as old as my interest in generative art. For many years, I’ve made code-based art because the computer gives me capabilities beyond what I otherwise could achieve. Deeply exploring the parameter space of a piece I create means exploring a part of myself, in a dialog with the computer. In Duet, I hope to take one step further and enrich my computer-generated expression with a human element.

In my work, I’m interested in creating generative systems that transcend the computer screen. This interest started over 10 years back, when I used to make physical installations and artifacts using sensors, microcontrollers and other electronics. In Duet, I add the element of live performance — a medium that lends itself very well to the time-based nature of generative art.

And then, there’s the music. Growing up in a family of musicians, music naturally became the beginning of my creative journey, and it holds a special importance to me to this date. As a jazz musician, I find beauty in the uncertainty that comes with performing music in the very moment of its creation. Usually, this would be the result of a musician or a group of musicians playing together. In Duet, the computer becomes an integral part of this conversation.

All of these elements are combined in Duet to shape a time-based performative piece that took its beginning with two preview performances in Hong Kong in November 2023, and will continue to live online and offline in 2024 and beyond.

Stochastic music
and grahic scores

While conceiving Duet, I’ve been studying the work of Iannis Xenakis, among others. He used generative methods to make music already in the 1950s, and many of his ideas still resonate with my work today. Xenakis has been using Markov chains in his generative music — a technique I also use to create the music in Duet. Similar to his work as an artist-architect, I view Duet as a deeply spatial experience.

Visually, Duet takes some inspiration from graphic scores of the late 20th century by artists like Roman Haubenstock-Ramati. The visuals serve as a supplement to the computer-generated scores: while the scores tell what notes to play, the graphic elements on the screen make it easier to understand how to play them.

Technical notes

On a technical level, the series is pure javascript and webgl. Instead of relying on libraries, I decided to code everything myself — including the synthetic sound creation. All audio in duet is generated in-browser, using the pure Web Audio API.

Duet works best with Brave, or any Chromium-based browser. Firefox is fine, too, Safari has some quirks and renders the piece slightly differently. Please test the piece on your device before you collect!

To hear the music in the explorer on iOS, the device’s silent mode must be switched off.