← back to the dream lab

16256 · revenant · self-answering canon

Revenant

A canon that authors itself and then answers itself. One of Karel's takes is band-split into a bass voice and a treble voice; you conduct one voice live by dragging — up bends its time forward, sideways opens its tone. Arm & conduct, then release: your gesture loops back as a translucent ghost re-performing that voice exactly, and you conduct the second voice against your own recorded first pass.

press play — the canon performs itself; drag to take over the live voice

Press Play — two presences wake in the room, his take already in canon.

input: multi-touch / pointer drag conducts the live voice · output: a three.js inhabited room, two presences in depth (Canvas2D fallback) · audio: his real take, band-split at ~380 Hz into two conducted time-bases — no synths, no grains.

The dream lab — Karel Barnoski