SPECTRAsound into form
THE CANVAS

A geometry of sound.

Ink above -30 dBFS
PERSPECTIVE / 3DDEMO COMPOSITION
Drag to orbit · scroll to zoomSwipe to scroll · pinch to zoom
LOW NOTES
HIGH NOTES55 Hz — 2.1 kHz
♫
Prismatic driftOriginal synth study · 120 BPM · 4/4
DEMO
0:000:32
Play & paint
BAR 01● ○ ○ ○

Playlist 0 songs

Drop songs anywhere to add them. Your queue stays in this session; audio never uploads.

    THE TRANSLATION

    A motif made from music.

    Resonance combines circular waves into a symmetric portrait for each estimated voice. The sound supplies the coefficients. There is no random seed or picture chosen from a filename. The original portraits remain available alongside supported-plate modes.

    Choose a cymatic membrane

    Original portrait preserves the previous Resonance drawing. Nine new membrane patterns combine circular vibration modes using Bessel functions. Nodes trace zero-amplitude lines. Antinode bands surround the strongest vibration lobes; Both overlays them. Circular symmetry ranges from 1-fold to 24-fold. One-fold uses a single mirrored motif with no extra rotational copies.

    Square and rectangular plate modes

    The six supported-plate options use the sine-product eigenfunctions of an ideal thin plate with simply supported edges. Square pairs combine (m,n) and (n,m), which have the same natural frequency. Music can change their balance without bending the plate coordinates. Balanced reference uses equal weights; the rectangular grid uses one mode. The drawing layers musical ink onto these valid spatial modes; it is not a time-domain simulation of a physically driven plate.

    Free edges, clamped edges, center supports, material, dimensions, and thickness change real Chladni patterns and resonant frequencies. A musical note has no universal cymatic picture. The provided frequency charts cannot calibrate a plate without those details. Antinode bands show high-amplitude contours, not exact maximum points. Show plate boundary reveals the ideal support edges.

    The song still supplies its signature

    Each membrane is bent by the same pitch, timing, register, and phrase features that shape your original portrait. The same pattern is not a fixed picture: different music supplies different curves, colors, and gaps. Earlier phrases occupy inner bands. Gesture tracking, legacy Harmonic symmetry, and Free pen have been retired from the studio to focus on these musical portraits.

    Pitch becomes a family of curves

    Continuous pitch is measured on a logarithmic musical scale. A Fourier transform combines pitch class with its position in the phrase, retaining note order instead of simply counting notes. Register changes size and color. Measured pitch variation adds fine ripples. Each voice's accumulated motif holds the picture together while individual phrases add variations.

    Why it is symmetrical

    The radius is a sum of cosine waves repeated around a circle. A wedge is rotated and reflected between 1 and 24 times, so the geometry and its painted gaps have exact rotational and mirror symmetry in 2D. This is an artistic mathematical mapping, inspired by wave interference—not a calibrated simulation of sand, a vibrating plate, or a claim that music is inherently symmetric. The membrane patterns are artistic deformations of standing-wave fields. Mixed modes and antinode bands illustrate amplitude structure, not a measurement of a real material.

    Time, rhythm, and silence

    Tempo and meter define the phrase clock. In 2D, earlier phrases sit nearer the center. In 3D they also move through depth. Each wedge follows the phrase's measured intensity: quiet intervals leave gaps. Live contours evolve from score frames up to the playhead, within the preanalyzed voice groups, so earlier contours can reshape as the voice's motif develops. Unpitched texture makes symmetric grains rather than invented notes. 4D is an artistic projection; perspective views need not look symmetric on screen.

    One intensity scale

    The same dBFS threshold applies to every recording and voice. Shape coefficients are normalized for legibility; loudness and the ink gate are not normalized per song. Listening volume does not change the image. File mastering levels still affect the result.

    Compare and keep

    Combined overlays the voices. Separated gives each its own portrait; Only isolates one visually. Playback remains the full mix. Export PNG prepares the finished track at 3000 pixels wide with your selected pattern, paint region, voices, and settings, even when playback is paused halfway through. Download PNG saves that image.

    These are estimated pitch and timbre groups, not separated instrument stems. A compact artistic image cannot uniquely identify every possible recording. The same audio and settings give the same result. Audio analysis stays on your device.

    Background: circular vibration modes · plate modes and edge conditions · musical Fourier analysis

    YOUR SOUND, CAPTURED

    Keep this composition.

    Your exported spectral sound painting
    ↓ Download PNG

    If your browser does not start a download, open or save the image using its image menu.

    SPECTRA, ON YOUR DEVICE

    Install your studio.

    iPhone / iPad: open this site in Safari, tap Share, then Add to Home Screen. Turn on “Open as Web App” if offered.

    Android: open in Chrome, then use its menu → Add to home screen → Install.

    Desktop: use the browser’s install icon or menu. The Windows EXE also supports direct speaker-output listening.

    This is an installable web app, with offline access after its first online visit. Music and background images are session-local; choose them again when reopening. Keep Spectra visible for playback and live capture. Device-audio sharing depends on the platform.