Recorded water × machine music · Vancouver, BC
Hush now,
little
supercluster.
A lullaby for a restless machine.
The water stays awake.
Encounter · A softer pulse, warm answering tones, and a closer stereo space.
Average level over time · each movement scaled separately
Settle in. Headphones welcome.
Download MP3Follow the current
Choose a moment to listen.
The lullaby / 07
Put the machine
to bed.
A small tune it can remember. A rhythm it can settle into. With every return, the machine asks for less of the room.
A seven-note lullaby returns over recorded river water. Warm key tones and a rocking accompaniment grow slower and quieter, leaving the final twenty seconds to the river.
WaterMachineSame vertical scale · average level over time
Sleep, Little Machine · Full mix · 04:00
Solo a layer without losing your place. Solo levels stay exactly where they sit in the mix; the water may be quiet.
Why did River and Coast look so similar?
They began with the same seven-minute machine arrangement. The newer mixes change its tone, melody and space, but keep its broad timing. The main player’s bars also average 1.75 seconds at a time and scale each recording separately. They show the rise and fall of level, not all the detail you hear.
These studies start with new scores. The two traces above separate water from machine on a shared scale, so you can see how much room each occupies. A similar curve can still contain very different sounds.
New rules, same real water
Sleep, Little Machine brings back the original C–E–G–B–A–D–C phrase. Seven returns grow slower and quieter. Rounded synthesized key tones, gentle rocking figures and sparse sustained harmony sit beside Thimras’s real Adelaide river recording. There are no sampled lullabies or voices. The last twenty seconds contain only river.
Relay Garden pairs synthesized metallic resonances with Thimras’s Adelaide river recording. Estuary Radio uses Joseph Sardin’s Dellec coast recording beneath detuned tones and quiet synthetic tuning noise. No broadcasts or voices are sampled.
A Machine That Waits uses Sardin’s outdoor rain recording. Its score measures rain level in five-second windows, then admits a new note only in a quieter window, at least ten seconds after the previous entrance. The rain also changes the machine’s level. This is an offline composition using the recording, not a live listening system.
The CC0 recordists and source links are credited below. Water remains at its original speed and stereo image, with level, DC correction and edge fades only. Every machine sound in these studies is newly synthesized. OpenAI Codex assisted with the scores, audio code and page. These are exploratory drafts, separate from the four-movement suite, awaiting Kris’s listening response.
The experiment
A lullaby for
a machine that
wants everything.
A sound experiment by Kris Krüg
The machine can be beautiful. It can also go to sleep.
The lullaby is the heart of this work: tenderness toward something restless, and a request that it stop taking up so much room. The newest study gives the machine a tune to return to, then lets it slow down, grow quiet, and rest while the water carries on.
This piece began with a protest sign: server racks tucked under a quilt, being told to hush. Made for Vancouver’s 2026 data-centre protest, the image becomes a four-movement experiment in sound and attention.
River and Coast open the suite with the two recordings Kris chose to keep. Weather Room brings rain into a slower field of overlapping harmony. Water After Us stretches the lullaby across eight minutes, lets coast become river, and leaves the final minute entirely to water.
Water keeps its own time. The machine has a score. The arc moves from encounter to expansion, shelter, and release. River keeps a softened pulse. Coast answers with a different melody and a more open space. In the last two movements, slowly moving voices and longer silences carry the music toward its ending. These are working drafts of a 28-minute suite.
See the protest art that started itWhat you’re hearing +
All water and rain comes from real field recordings. The musical parts are synthesized. The refined mixes add answering melodies, slow stereo movement and diffuse reflections to the machine parts. The water keeps its recorded stereo image. Earlier drafts remain available for comparison at the same listening position.
Full movements play as 320 kbps stereo MP3s. The featured lullaby also has a full 24-bit / 48 kHz WAV download. The mixes are close in measured loudness. This is two-channel stereo, designed with headphones and stereo speakers in mind. The page applies no additional audio processing. Tracks have separate fades; playlist transitions are not a gapless album master.
Water retains its original speed and pitch. This refinement reuses the previous field-recording edits and changes their level only. Those edits consist of DC correction, fades and crossfades. There is no synthesized water, water EQ, denoising, compression, or added reverb. In the four-movement suite, musical timing is composed, not driven by environmental measurements. The separate listening laboratory explores a rain-responsive score. The machine-only reverberation ends before each water-only closing passage. Final artistic review remains in progress.
Place, process & credits +
The political starting point is Vancouver, on the unceded traditional territories of the Musqueam, Squamish, and Tsleil-Waututh Nations. The river and coast recordings come from Adelaide and Dellec beach. The rain recordist does not specify a location on the source page. These are not presented as BC field recordings. This work does not represent Indigenous knowledge.
I / River: Park ambiences by Thimras, recorded in a public park in Adelaide, South Australia. Original stereo WAV: 16-bit / 48 kHz. The source description includes birds, frogs, and other wildlife.
II / Coast: Dellec beach #4 by Joseph Sardin / BigSoundBank, recorded near Brest, France. Original stereo WAV: 24-bit / 48 kHz. Both recordings are released under CC0 1.0.
III / Weather Room: Rain and Thunder #4 by Joseph Sardin / BigSoundBank, a CC0 outdoor recording in 24-bit / 48 kHz stereo. This draft uses one continuous passage, from 27:30 to 33:30. “Weather Room” describes the imagined listening space.
IV / Water After Us: The credited coast and river recordings meet in a 90-second crossfade. This is an imagined geography. The machine is silent for the final minute.
The project grows from Kris Krüg’s environmental photography and data-centre protest art. OpenAI Codex assisted with research, concepts, composition and synthesis code, writing, and this listening room. The accompanying concept artwork is AI-generated. The audio is rendered with conventional digital signal processing.
You Can’t Drink Data · Kris’s photography · Territorial context