A field, not a playlist
Every sound has a position in three dimensions. Distance sets level and filtering, height changes the ear filters, and the head-related transfer function does the rest.
Spatial audio studio · runs in the browser
Put sound where you want it. Then move it.
Drop birdsong two metres to your left and four above. Send a whale call arcing across the field over ninety seconds. Lay a ten hertz binaural beat underneath. Everything renders locally through the Web Audio API, with no account, no upload and no tracking.
- Sounds and tools
- 62
- Frequency tools
- 16
- Built-in journeys
- 9
- Account needed
- no
field view · top-down · distances in metres
A timeline, not a loop
Clips say when a sound plays. Keyframes carry position and level through time, so a journey has an arc instead of a constant bed.
A focus view, not a studio
When you only want frequencies, the whole studio collapses into one screen: a mode, a length, one button and the numbers that matter.
01 · The field
Spatial audio you can see
The field is a metre-accurate map of the space around your head. Drag a sound to move it on the ground plane, hold Alt to lift it, and switch to the tilted 3D view to see height directly.
- Binaural rendering. Each source runs through a head-related transfer function panner, plus modelled shoulder reflection and outer-ear filtering that shift with elevation.
- Posture. Standing, lying on your back, lying on your side. The listener orientation and the anatomy filters change together.
- One camera, two views. The 2D map and the 3D orbit are the same scene blended between orthographic and perspective, so nothing jumps when you switch.
- Audio-reactive nodes. Each source has its own analyser. The halo and ripples follow the actual signal, not a random timer.
- Motion paths. Orbit, ping-pong, drift and breathe run without a single keyframe, and show their path on the map.
- Other outputs. Stereo speakers, 5.1, or a custom speaker layout you place by dragging.
- Head tracking. On a phone, device orientation rotates the listener so the field stays put when you turn your head.
02 · The timeline
Sessions with a shape
A journey is a set of tracks over a fixed length. Every sound gets a clip and, if you want, a path. Named sections mark the phases so a forty-five minute session reads at a glance.
- Clips. Drag to move, drag the edges to trim. A source fades in when its clip opens and out when it closes.
- Keyframes. Drag sideways to change the time, up and down to change the level. The map draws the resulting path.
- Mute and solo per track, so you can hear one layer against the bed while you build.
- Snapping that follows the zoom: half a second when you are close in, thirty when you are looking at the whole hour.
- Undo everywhere. Moves, adds, deletes, clip edits, keyframe edits, parameter changes and motion changes all sit on one stack.
- Nine journeys built in, from a twenty-five minute deep-work block to an hour-long sound bath and a thirty minute sleep descent that ends in silence.
03 · Focus view
One screen when that is all you need
Not every session needs a map. The focus view is a single player: pick a mode, pick a length, press the ring. The readout shows exactly what is playing, in hertz. Sessions have a settling phase, a main phase and a wind-down, and a calming session tapers its own frequencies on the way out instead of just going quiet.
Focus
10 Hz binaural · pink noise
Twenty-five minutes of speech-shaped masking with a beat underneath.
Deep work
brown noise · pad
Fifty minutes, no beat, nothing to follow.
Calm
6 breaths/min · 6 Hz
A breathing pacer at resonance rate with a theta beat.
Sleep
2.5 Hz · brown noise
Forty-five minutes that fade to silence over the last three.
Clarity
40 Hz isochronic
Twenty minutes of gamma-rate pulsing over a quiet drone.
Single tone
432 Hz, adjustable
One frequency you set yourself, with the solfeggio set one tap away.
04 · The library
Forty sounds, seven categories
Recorded ambience for place, synthesised material for texture, and procedural generators for anything that has to hold a frequency exactly.
| Category | What is in it | How it is made |
|---|---|---|
| Nature | Birdsong, rain, thunder, waves, crickets, campfire, jungle at night, river, monkeys, elephant, leopard, deep ocean, whales, dolphins | Recorded samples, looped |
| Places | Café, train, city traffic, city park, subway station | Recorded samples, looped |
| Sound healing | Singing bowls, three gongs, wind chimes, a seven-note chakra bowl set | Samples, chakra notes pitch-shifted |
| Instruments | Piano, synth pad, bass, strings, flute, bells, drone, arpeggio | Rendered offline from oscillators |
| Frequencies | Binaural, isochronic and monaural beats, pure and solfeggio tones, white, pink and brown noise, a breath pacer, a 7.83 Hz pulse | Procedural, exact |
| Generative | Evolving chord pad, deep drone, sparse shimmer | Procedural, never repeats |
| Your files | Any MP3, WAV, OGG or M4A you drop in | Decoded in the browser, never uploaded |
Audition any card before you place it. Uploads stay on your machine: there is no server to send them to.
05 · Frequency tools
Labelled for what the research actually says
Plenty of apps sell frequencies with claims the literature does not support. This one ships the tools and tells you where the evidence stands, on the card and in the inspector.
| Tool | What it does | Label |
|---|---|---|
| Pink and white noise | Broadband masking. Pink rolls off the top end, so it sits behind work instead of over it. | Some evidence |
| Brown noise | Deeper still, and popular, but no trial covers this colour specifically. Treat it as masking and texture. | Weak evidence |
| Breath pacer | A swell you breathe with, default six breaths a minute. | Some evidence |
| Binaural beats | A different frequency in each ear; the beat is the difference. Needs headphones and stays locked to your head. | Weak evidence |
| Isochronic and monaural beats | A pulsed or summed tone that works on speakers too. | Weak evidence |
| Solfeggio tones, 432 Hz, 7.83 Hz | Popular frequencies, offered because people ask for them. | No evidence |
| 40 Hz gamma | Clinical work on 40 Hz uses light and sound together, daily, with diagnosed patients. That does not transfer to listening here, and the app says so on the card. | Weak evidence |
Each label carries a one-line reason, not just a verdict. Pick a binaural beat and the inspector points out that it needs headphones, and offers to swap it for an isochronic pulse that works on speakers. Full references live in the repository under docs/research.
06 · Under the hood
Plain JavaScript, no framework
Audio
Web Audio API. HRTF panner per source, shoulder and pinna peaking filters, low-pass and high-pass inserts, tremolo, a shared convolution reverb bus and a master limiter.
Binaural
Beats bypass the panner and merge straight to left and right, because spatialising them would destroy the beat. The app says so in the inspector.
Rendering
Canvas 2D with a real camera: yaw, pitch, and a blend between orthographic and perspective. Picking and dragging run through the inverse projection, so they stay exact in every view.
State
A command stack for undo, scenes in local storage, and shareable links that carry the whole journey in the URL.
Build
Vite. Two pages: this one and the app. Deployed to Vercel's edge.
Tests
Vitest with jsdom across the audio engine, generators, camera, timeline, presets, scenes and the undo stack.
Open it and place a sound.
Headphones give you the full effect. Nothing installs, nothing signs up.
Open the app