pressure
Average signal energy determines how deeply the artifact appears compressed into the field.
signal / live study
SOUNDFOSSILS
What shape would a sound leave if sound could fossilize?
Let the browser listen for a few seconds. Energy, spectral balance, roughness, crossings, and transients become strata, pressure, pores, and fractures. The audio itself is never recorded or uploaded. Only the measurements survive as form.
listen to a moment · make a sound · stop when the material feels complete
no fossil yet
energy00%
brightness00%
roughness00%
fractures00
duration0.0s
Microphone access is requested only after you choose to listen. SOUND FOSSILS does not use MediaRecorder, create an audio file, or send microphone data to Warp the Surface. The live stream is stopped as soon as you stop listening. If microphone access is unavailable, the demo path remains fully playable.
under the surface
The sound disappears. Its behavior remains.
strata
Capture duration and changing energy leave nested layers rather than a literal timeline or waveform.
porosity
Spectral flatness and zero-crossing activity create pores and grain, making noisy sounds materially different from tonal ones.
fracture
Sudden energy changes become fault lines. A clap, impact, or consonant can split an otherwise smooth deposit.
system
Analysis is temporary. Form is the memory.
- input
- microphone / explicit browser permission
- analysis
- Web Audio API / AnalyserNode
- features
- RMS · spectral centroid · flatness · zero crossings · transients
- rendering
- Canvas 2D sediment + fracture field
- recording
- none
- server inference
- none
- status
- experimental / mobile-first v0
This first build intentionally avoids an audio model. Native signal features already provide enough structure to test the artistic question with lower latency, no model download, and no inference cost. A later local model should be added only if semantic listening creates a genuinely different material.