How it's made
Frekans Studio is a sound laboratory: every sound is generated on this device, and every piece of content shows its level of evidence. This page explains how it works, who made it and what has changed.
Manifesto
- The sound is made right here. Every tone, every beat, every raindrop is computed on this device, right now; no audio files are streamed from a server.
- We say what we know, and what we don't. Every session shows its level of evidence; we never present weak evidence as strong.
- Listening is an experiment. You can see the sound, take it apart layer by layer and rebuild it in your own room.
- Your ears come first. Sound fades in gently and never jumps; nothing on screen flashes more than three times a second.
- No account, no tracking. What you listen to stays on your device.
One engine, on your device
Every sound comes from a single signal-processing core. In the browser it runs in real time in an AudioWorklet; file export uses the same core, and the phone app renders each session with that same core ahead of time and plays it with the device's own player.
The library holds 159 sessions. Each is described in code and a seed; none of the sounds is a recording. Rain grows out of a droplet model, a bowl out of its modes, noise out of its colours.
- DC blockerFilters out any zero-frequency offset.
- 20 kHz band limitCuts everything above the audible band.
- Envelope × masterGentle fade-in and fade-out; the level rises by at most 6 dB per second.
- True-peak limiterThe output never exceeds −1 dBFS.
A field that sees the sound
The Resonance Field draws the layers of the sound as rings and waves. It is tiered to the device's power: a still engraving, a 2D canvas or WebGL2. In browsers that support it, the heavy drawing runs on a separate thread and never touches the timing of the sound.
The picture never changes at the rate of a beat and never flashes more than three times a second (WCAG 2.3.1); every new visual passes an automated flash and pattern test. With reduced motion the field becomes a still engraving, and the “Pause visuals” button stops it everywhere.
How we grade evidence
Every scientific claim gets one of six levels: ● A · ◆ B · ▲ C · ■ D · ✦ T · ○ S. A session is never shown as stronger than the weakest claim it links to. The sentences follow the certainty language of the GRADE Working Group and the Cochrane Handbook: “probably”, “may”, “we don't know”.
Today 34 claims and 90 sources are shown. Every citation was checked against its Crossref or PubMed record; every claim has a version and a changelog. Evidence methodology →
No account, no tracking
You never create an account; the app contains no analytics, advertising or tracking software. Your listening history, notes and settings are kept only on this device, in the browser's storage; you can export or delete them in Settings.
The microphone opens only on tool pages, when you press the button; nothing is recorded and nothing is sent anywhere.
Credits
- Design, sound and code
- Murat Tunalı
- AI partner in development
- Claude (Anthropic)
- Typefaces
- Imbue (Etcetera Type Co.), Instrument Sans and JetBrains Mono, all three under the SIL Open Font License 1.1.
- Software
- React, Next.js, Capacitor, Zod, Lenis and GSAP. The MP3 encoder is LAME (LGPL), loaded as a separate module.
- Evidence sources
- 90 publications; citations from Crossref and PubMed records. Evidence library →
The role of AI in drafting the claim texts, and the verification steps, are set out in the evidence methodology. AI and review →
Changelog
Version 1.0
- How it's made: a manifesto, how the sound and the picture are produced, credits and this changelog.
- A mini listening room on the home page; three short lessons in the room on direction, level and echo.
- A “Listen as you roam” switch on the Atlas, and sound words under its regions.
- A new bottom bar on phones: Listen · Atlas · Room · Evidence · More. The sound design space is now called the Listening room.
- From the Atlas to a session in under a second; empty and idle states redone.
- A lighter first load: the app shell's JavaScript is under 200 KB.
- A share preview on every page; larger touch targets on phones and tablets.
frekans.studio goes live
- The listening room: place sounds around your head; direction and distance change the sound itself.
- Journey pages, guides and frequency articles, a redesigned evidence library and claim pages.
- Works offline; clear messages that say why when sound cannot start.
Atlas, Lens and the living session page
- Sonic Atlas: the library as a map drawn from the measured similarity of the sounds.
- Frequency Lens: see sound as wave, spectrum, stereo and time in the Experiment.
- Session page: how the sound is made, comparison with another sound, the evidence lens; one question when a session ends.
Resonance, silence and the Resonance Field
- Resonance Lab: play the bowl, the bell and the gong by hand; Intervals and the Sound → Silence timer.
- Resonance Field: a visual field that draws the sound live, tiered to the device's power.
- A dark primary theme, keyboard shortcuts and a search palette (Cmd/Ctrl+K).
New sound engines and previews
- Twelve new sound engines and 52 new sessions: nature, places, resonance, breath and generative focus.
- Preview any session for 60 seconds; gentle crossfades between sounds.
- Session Finder, A/B comparison and a fresh variation of the same recipe.
Library and sound design
- A new library: 86 sessions, shelves and filters; water, bowl and rain sounds rebuilt from scratch.
- Modulation, effects, macros, generative recipes and microtonal tuning in the sound design space.
- First publication review of the evidence library: 21 claims published.
Web and phone
- The web app and the phone shell share one common layer; cluster drones and cinematic layers.
- The first sound design space: a one-screen instrument and pads.
First version
- The sound engine: binaural, isochronic and coloured noise, generated on the device.
- The evidence database and a checker that catches health-claim language.