SONIKOSMOS: SOL-ω
The instantaneous angular velocity of our solar system, made audible.
Nine voices — the eight planets, each one driven by its actual instantaneous angular velocity right now, pulled live from NASA JPL's DE441 ephemeris (the same model used for real spacecraft navigation), plus the Sun's own real rotation rate. Not an average. Not a guess. Not a pleasant approximation built to sound good. What you hear is not designed to sound pretty. It is designed to be true.
When the solar system forms rare JI chords, mode convergences, and clean intervals — precomputed across all of 2026.
Loading…
Everything above plays itself. SPHAERA is the real-time performance instrument built on the same live orbital data: plug in a MIDI controller and play the actual solar system with your own hands — nine bodies, nine fixed keys, each one tuned to true, unquantized orbital pitch the instant you press it.
Every voice's real time-domain signal, drawn on top of each other. This is what "nine microtones close together" actually looks like as raw waveform — watch the combined shape shift as they drift in and out of phase.
Real-time XY parametric plot: horizontal = one planet's waveform sample, vertical = another's. Simple integer ratios trace stable geometric shapes — a perfect octave (2:1) draws a figure-8, a perfect fifth (3:2) draws a pretzel. The closer two voices are to a clean ratio, the more the figure holds still. Our planets are microtonal; the figure drifts slowly and rarely locks. Pick any two voices above to compare them.
The standard idealized free-square-plate model F(x,y) = cos(nπx)cos(mπy) − cos(mπx)cos(nπy) — bright where a real plate wouldn't move (nodes, where sand collects), dark where it would (antinodes). Each voice's mode pair comes from its real pitch rank; each contributes its real instantaneous sample and real mix weight, so muted/solo'd-out voices drop out of the figure exactly as they drop out of the sound.
Music Analysis
The actual chord, right now, written the way a musician would have to write it: nine noteheads, each one placed at its nearest equal-tempered staff position, with how far it really sits from that position printed right next to it. Nothing below is quantized — the staff position is a reference grid, the cents number is the truth.
Mercury and Neptune sit ~9 octaves apart, which would mean dozens of ledger lines if every note were drawn at its true register — unreadable. Instead, each note is folded to the nearest octave of the same letter near the staff's center, exactly the way real scores use 8va/8vb (an octave higher/lower than written) or 15ma/15mb (two octaves) instead of stacking ledger lines forever — printed in amber above each note. Beyond two octaves there's no standard symbol, so further notes are marked plainly ("+6 oct", etc.) rather than inventing a fake-authoritative one. Where the note sits left-to-right still reflects true pitch order, low to high; where it sits on the staff plus its octave marker is what it actually sounds like. This shows all nine voices regardless of any mute/solo currently active above — it's the chord these nine bodies are actually forming right now, not just whatever subset you happen to be listening to. Updates with every refresh and every pitch change.
Reading the amber labels above each notehead: 8va/15ma = this notehead is drawn lower than it really sounds — the real pitch is 1/2 octave(s) higher than written. 8vb/15mb = the opposite: the real pitch is 1/2 octave(s) lower than written ("va" = alta, higher; "vb"/"mb" = bassa, lower — standard Italian ottava terms). Beyond two octaves there's no standard symbol, so it's just printed as plain text ("+N oct"). Every label is the true register that planet's voice is actually in — folding only changes where it's drawn, never the audio.
All 28 pairwise intervals between the 8 planets, right now, in one view. Each line's color and weight encodes consonance: bright green = simple JI ratio (perfect 5th, major 3rd…), dark red = complex dissonant ratio. Hover any line for the exact interval. Planet labels show current sounding note and cents deviation. Updates with every data refresh.
All nine voices folded into a single octave and sorted — what this chord would be if register didn't exist. Each voice's real pitch, named as a note plus how far it actually sits from that note, in cents.
| Body | True note | True freq | PC | Cents |
|---|
What's notable right now: chord possibilities ranked by how many voices fit each shape, just-intonation singularities when any two planets land near a simple integer ratio, and other harmonic events in the moment.
The same pitch-class cluster above, replotted in the standard fifths arrangement (each step a real perfect fifth) instead of chromatic order — this is the actual lens tonal/jazz harmony uses to reason about chord and key relationships. Filled dots are active pitch classes, colored by voice; line brightness between two active dots reflects real fifths-distance, not style. The dashed ring marks the matched chord's root. If two or more planets land on the exact same pitch class, that dot splits into one wedge per voice with a small count badge — real coincidences are shown, never hidden behind one flat color. Each active pitch class is labeled with which planet(s) sit there and their exact cents deviation — the wheel position is the nearest 12-TET reference, not the truth; the cents number next to each name is the truth. No new data here, just a different read of the same notes.
Every chart above places these 9 voices on a 12-TET grid, because that's the only way to give a frequency a conventional name — but the grid is a reference, not the truth, which is why cents-from-grid is printed everywhere next to it. This chart drops the grid entirely. Each planet sits at the exact, continuous angle and radius given by log2(its real frequency ÷ the lowest real frequency currently playing) — nothing is rounded to any scale, just intonation included. One full turn is exactly one real octave; the same pitch class an octave higher lands strictly farther from center, never on top of the lower one — the same model as Shepard's "helix of pitch." The dashed spokes mark a few real just-intonation ratios (3:2, 4:3, 5:4, 6:5, 5:3) purely so you can compare an angle against them by eye; nothing is snapped to them. The amber path just connects the voices in true pitch order, so whatever real shape they trace through register-and-chroma space is visible as a shape, not a list of numbers. The reference frequency (the t=0 anchor, marked with the amber ring) isn't a fixed constant — pick which voice it is below; it only changes which frequency everything else is measured FROM, never the real frequencies themselves.
- Center / innermost ring — the reference voice's own frequency. It's only an anchor point you picked above, not a special or "correct" pitch.
- Each ring further out — exactly one real octave (a doubling of frequency) farther from the reference. Crossing one ring always means the frequency exactly doubled; nothing else does.
- Angle around the circle (chroma) — the position of a pitch within one octave, continuously, never snapped to 12 notes. This is the only thing deciding whether a dot ends up left, right, up, or down — there's no "up = higher pitch" rule the way a staff has. "Up" only means "the same chroma as the reference voice" (angle 0, by convention); every other direction is simply a different chroma, wrapping around like hours on a clock face.
- Two dots at the same angle, different radius — the exact same pitch class, a different octave. They sit on the same straight line from the center.
- Two dots close together in angle — close in pitch "color," regardless of how far apart their registers are.
- Dashed spokes — real just-intonation ratios (3:2, 4:3, 5:4, 6:5, 5:3), marked only so you can compare a dot's angle against them by eye. Nothing is forced onto them.
- Amber path — just connects the voices in true pitch order so whatever shape they form is easier to trace. It carries no musical meaning on its own.
Same pitch class (A), four real octaves apart. All four land at exactly the same angle — only the radius changes. This is what "the same note, higher" looks like here.
A literal just-intonation major triad (4:5:6): a root, a true 5:4 major third, and a true 3:2 perfect fifth. The third and fifth land exactly ON their matching dashed spokes — not a coincidence, that's what makes them "in tune" in the just sense. The triangle is what a clean major triad looks like here.
Three voices only 2–5 Hz apart — a tone cluster (check the Hz labels, that's the real truth). The chart always stretches whatever span is actually present to fill its full radius, so with nothing else around to set the scale, this tiny gap still spreads across the whole circle. In the live 9-voice chart above, the same three voices would instead collapse into a tight little knot, because Mercury-to-Neptune's much bigger spread sets a much bigger scale. The angle here is still exact either way — only how "zoomed in" the radius looks changes.
What this actually is
The data: by default, every 60 seconds, this page asks NASA JPL's Horizons system — the Solar System Dynamics group's ephemeris service, built on the DE441 model that real spacecraft use for real navigation — for the exact position and velocity of each of the eight planets at this exact moment. Not a table of averages. Not an approximation. The same numbers a mission planner would pull to point a probe. The same query can also be aimed at a different moment entirely — click the clock above the 3D view to pick one.
The physics: each planet's frequency is ω(t) / 2π, where ω = |r×v| / |r|² — Kepler's Second Law, computed directly from those real position and velocity vectors. A planet near perihelion is genuinely sweeping faster, and genuinely playing a higher note, than the same planet near aphelion. That variation is real physics, not a sound effect layered on top.
The Sun is different and is labeled as such throughout: it doesn't orbit anything, so there's no live ω to query for it. Its frequency comes instead from its known equatorial sidereal rotation period (~24.47 days) — a literature constant, not a live Horizons number, because the Sun rotates differentially (faster at the equator, slower near the poles) and has no single honest "rotation rate" the way a planet has a single orbital ω at a given instant.
The tuning rule: there are actually two octave multipliers stacked on top of each other, and they behave differently on purpose. The first is automatic: an octave multiplier locked once, the instant the engine starts, by dragging Earth's literally-inaudible real frequency up near 110Hz — not yet locked. It never recalculates afterward, including if you later jump to a different moment with the date control, because relocking it would change the apparent pitch between sessions for no physical reason. The second is the pitch slider you control directly (0 to +8 octaves, in the Engine & Mix panel) — it's intentionally a second, independent multiplier rather than folded into the first, and it's measured in whole octaves rather than semitones, because a semitone step assumes 12-tone equal temperament — the very reference grid the note readouts elsewhere measure real frequencies against, never assume. Mercury–Neptune span ~8.96 octaves of real, fixed ratio; human hearing spans ~9.97 — so try roughly +4 to +5 on the slider to bring Neptune up near 20–30Hz without pushing Mercury past 16kHz. Whichever multiplier is doing the work, both of them are applied identically to all nine voices — that's what keeps the relationship between every pair of voices mathematically exact; transposing each one by a different number of octaves (a common shortcut in other "music of the spheres" projects) would quietly destroy the real ratios between them.
The per-planet re-octave control on each voice card is a third, independent layer — whole octaves only, applied only to that one voice, so it shifts one voice's register without touching any other. "Base note / base hz" on each card shows what the engine placed that voice at (before your per-planet shift); "note / hz" shows where it actually sits after your shift. When the per-planet slider is at 0, both readouts are identical.
Honest expectation: unlike resonance-chain exoplanet systems (TRAPPIST-1, TOI-178), our own eight planets are not in clean integer-ratio resonance with each other. What you're hearing is closer to nine close, slowly drifting microtones than a chord — and that dissonance is the actual finding, not a flaw in the synthesis.
A Brief History of Trying to Hear This
The idea that the planets' motion constitutes a form of music is at least 2,500 years old, and it was argued about from the start. Every entry below is a real attempt to answer the same question — read top to bottom, it's a lineage, not a list.
Pythagoras and his school proposed musica universalis — that each celestial body, moving at its own speed along its own orbit, produces a tone determined by that speed, and that the combined motion of all of them is a real, physical harmony. Pythagorean cosmology held that this harmony is inaudible to us not because it isn't real sound, but because we have heard it continuously since before birth — the mind filters out a sound with no silence to contrast it against.
Aristotle took the idea seriously enough to refute it directly. In On the Heavens, he argued that bodies that large, moving that fast, would have to produce an unimaginably loud sound — and since we plainly don't hear one, the Pythagorean picture had to be wrong. It's worth knowing this critique existed alongside the original idea for as long as it has: "the music of the spheres" was never universally accepted as literal, even in antiquity.
Ptolemy, in his Harmonics, kept the mathematical thread alive — connecting musical interval ratios to astronomical ones without necessarily claiming anyone could literally hear it, treating the correspondence as a structural fact about the cosmos rather than an acoustic event.
Boethius didn't add new astronomy — he preserved the idea by giving it formal structure. De Institutione Musica divides music into three kinds: musica instrumentalis (what you actually play), musica humana (the harmony of body and soul), and musica mundana — the music of the universe itself, real but inaudible to human ears. The text became required reading in medieval universities for close to a thousand years. Without that specific survival route, there's a real chance the Pythagorean idea doesn't reach the Renaissance intact enough for anyone to ever test it against real data.
Brahe never claimed the planets sing. What he did, over decades at his Uraniborg observatory, was compile the most precise naked-eye astronomical observations Europe had ever produced — measurements exact enough that small, real discrepancies in planetary motion couldn't be explained away anymore. Kepler inherited that exact dataset after Brahe's death and used it to compute the real perihelion-to-aphelion velocity ratios in Harmonices Mundi. The pattern repeats here precisely: someone has to build the precision first. Brahe is this project's NASA, four centuries early.
Kepler is the actual hinge. Using Tycho Brahe's observational data — the most precise available anywhere on Earth at the time — he computed the real ratio between each planet's angular velocity at perihelion and at aphelion, and assigned each one an actual musical interval based on that real, measured variation. This is not numerology layered after the fact: it's the same method this page uses, four centuries earlier, with the best real orbital data Kepler could get his hands on instead of a live ephemeris API. The instantaneous speeding-up-at-perihelion this engine sonifies is exactly the phenomenon Kepler was listening for.
The thread resumes in a new form: researchers and sonification artists — System Sounds' work on TRAPPIST-1, the ESO's sonification of TOI-178 — built real, data-driven audio straight from orbital data. The decisive move was choosing well: both systems sit in genuine near-integer orbital resonance, real ratios close enough to simple musical intervals that each one can be nudged — rounded — onto the nearest note of an actual scale. That rounding is the craft, not a shortcut: it's what turns real astronomy into something a listener immediately recognizes as a melody, a genuinely lovely piece of translation. It's also a different choice than the one this project makes below — tuned for what sounds pleasing, rather than left exactly where the unrounded math puts it.
Every entry above solved one of two problems but not both. Pythagoras, Ptolemy, even Kepler himself, had exactly the right question and no way to keep asking it — a calculation done once, by hand, frozen at the moment of publication. System Sounds and the ESO had the live-instrument fidelity, pointed deliberately at systems chosen because their orbits already resonate cleanly. Our own solar system was left alone precisely because it doesn't cooperate.
As far as this project's own research could establish, this is the first time both problems have been solved by the same instrument at once: a live connection to the same DE441 ephemeris that flies actual spacecraft, continuously recomputing Kepler's exact quantity — instantaneous angular velocity, not average period — for the one system nobody sonifies this way, because it's the one system that refuses to resolve into a chord.
Pythagoras guessed the harmony. Kepler calculated it once, by hand, and called it finished. This is that exact question, finally given a nervous system — answered continuously, live, for the solar system we actually live in, not a tidier one chosen in advance because it would behave. The lineage above didn't end. It arrived here.
"Once, by hand, frozen at the moment of publication" no longer has to mean once. The same live connection can also be pointed at any moment between 1700 and 2300 — so the exact configuration of the sky the night Harmonices Mundi was published is just as real, and just as available, as this instant.