The part of the work that looks forward: generative instruments built to measure, not software bought off a shelf. They come from a practical question — what happens if sound can draw, and if the image can answer the musician while they play.
AI-assisted production
Music videos made with generative pipelines: images and movement built on the track rather than cut over it. Suited to independent productions without a crew budget that still want an image of their own.
Contemporary and ambient music
Applications written to measure for experimental shows: every project has its own rules, and the instrument is built around the score rather than the other way round.
Audio and image answering each other
Systems that control and alter audio and video live, while the performance happens. Analysis of the incoming sound — frequency, voice, room — becomes a parameter: what you hear changes what you see, in the same instant.
This is not theory. These instruments are built by testing them on stage, inside our own work before anyone else's. Two have already been built and performed, and they run in opposite directions.
Application — from image to sound
Synaesthesia in reverse: the image is made to sound. The system reads colour — hue, brightness, density — and translates it into musical material, choosing instruments, scale and tuning through different generative algorithms. No clip library: what you hear comes from what you look at, and every pass produces a different outcome.
Application — from sound to image
The opposite path: sound becomes image. The application listens to what is happening in the room and turns frequency into colour and stroke — low and high produce different chromatic gestures, so every performance is an unrepeatable painting. In parallel, speech is recognised, written and processed as a graphic layer over the video. Connecting an instrument over MIDI is an extra option, not how it works: the sound that is already there is enough.