The questions, answered straight.

Soundscaper listens to what you need from the next twenty, sixty, or ninety minutes — or eight hours — and composes a personalised soundscape for it. Every session opens with a short headphone induction (a brief gateway phase using a Monroe-style binaural technique), then settles into a sustained body shaped by your intake. The audio is composed fresh on our own servers, not pulled from a back-catalogue. Two sessions are never identical. The science behind every design decision is tier-graded and public on the science page.

Calm and Headspace are content libraries — a finite shelf of recorded tracks, you pick one. Soundscaper writes a fresh session from your specific intake. The composition is built to your length, your context, your tolerance for binaural work, and the eight-or-so other choices you made at intake. It's a new session composed each time, not a track pulled from a shelf.

Closer cousins, with three structural differences. Your intake is read by a language model Kenjin self-hosts — never sent to a third-party AI API like OpenAI or Google, and never used to profile you. Every session opens with a Monroe-style headphone induction before the sustained body begins. And the same composition can be delivered through five different audible surfaces — nature, nature plus composed, soundscape music, ambient, or pure binaural — so the same underlying work doesn't have to sound the same.

About a minute. Six structural choices (context, length, headphones, intent, instrumentation, cues), one optional sentence about how you're actually feeling, and three safety confirmations. You can stop mid-flow and the system holds your place.

For the opening induction, yes — binaural work only does its job on headphones. If you'd rather not, the wizard offers a body-only composition with the induction removed. You confirm at intake; we don't guess.

Eleven, from twenty minutes to eight hours: 20, 40, 60, 90, 120 minutes, then 3, 4, 5, 6, 7, 8 hours. Anything four hours or longer is treated as overnight — a short gateway, a long sustain phase carrying you through the night, and a soft handoff at the close so you don't get yanked awake.

The same therapeutic technique stack can be delivered through five different audible textures: nature-first (water, forest, rain, wind), nature plus composed (the default), soundscape music, ambient music, or pure binaural. The technique stays constant; the room it lives in changes.

Yes. On a nature-first or nature-plus-composed surface, the wizard exposes a multi-select: water, rain, wind, forest, birdsong, distant thunder, fire crackle, insects. The composition pulls matching stems from a CC0-only library.

Not yet — it's a V2 release. When it ships it'll be headphone-only and optional, surfaced through a toggle in intake. We'll treat the comfort and immersion benefit as the durable effect; we won't claim spatial rendering deepens meditation.

Context — bedtime, meditation, focused rest, active work, background, care companion — reshapes tonal range, pacing, transitions, and how the session closes. A bedtime soundscape ends in a sleep handoff with no wake-up. A focused-rest soundscape brings you back to ordinary alertness.

A compressed FLAC or MP3 you can download and play anywhere — with the intro, provenance line, watermark announcements, midpoint cue, and five-minute closing cue baked in. From the Monthly Practice plan upward you also get clean, unwatermarked playback inside the app.

Hosted library access stays live for three months after you stop subscribing. Compressed downloads on your devices are yours regardless. After ninety days of lapsed subscription the master files become eligible for deletion — your downloads remain.

Personal use, yes — including listening with members of your household, and (on the Companion plan) with the second listener profile. Commercial distribution or sublicensing needs our permission. Soundscapes are not licensed as training data for other AI models.

No. Your first session is a one-off Single Session purchase, with no recurring charge — the price is confirmed when early access opens. Pricing applies when you generate, not when you sign up.

No. The plan is sized for a typical practice cadence. If you want more, Companion or pay-as-you-go credits cover it.

Yes — there's an early-access programme for clinical and care contexts. Email [email protected] with a short description of your use case.

GBP, USD, and EUR on the web. Mobile in-app purchases use App Store and Play Store pricing per platform policy.

On our own servers. Intake parsing, safety classification, composition planning, and audio composition all run on servers Kenjin operates — using a language model Kenjin self-hosts, never a third-party AI API. Our servers also hold your account, your subscription state, and — while you're subscribed — cloud copies of the master files. Your device holds the compressed downloads you save and plays them back.

Account fields (email, hashed password, optional display name), subscription state, recording metadata, journal entries, push tokens if you register them, and operational logs (request id, IP, status code) for uptime and abuse prevention. The full list lives on the data policy page.

No. Your intake is parsed into a structured profile by a language model Kenjin self-hosts on its own servers. Only the structured profile is stored with the session; your raw intake is not retained, and it is never sent to a third-party AI service.

No. The public site carries no third-party analytics. The signed-in member area uses a small number of strictly-necessary cookies for authentication and session continuity. There's no analytics consent banner because there's nothing to consent to.

The system blocks generation and routes you to the help page, which surfaces UK helpline routes (Samaritans 116 123, NHS 111, Mind) and the international Find a Helpline directory. Real human services. Soundscaper can't help in crisis; it shouldn't pretend to.

Anyone with photosensitive epilepsy or seizure history, pacemakers or implanted cardiac devices, current pregnancies where binaural work is contraindicated by your clinician, and anyone under 18. Vestibular and motion-sensitivity disorders warrant caution too. The safety page has the full list.

If you get headache, dizziness, nausea, or any unusual sensation that doesn't pass quickly, stop the session and check in with a clinician. Long-form quiet listening is generally well-tolerated; the audio engine holds conservative loudness ceilings (−16 LUFS general, −20 LUFS sleep) and a −1 dBFS true-peak ceiling.

Established — multiple high-quality RCTs, ideally synthesised in a Cochrane review or peer-reviewed meta-analysis. Emerging— single trials or small samples; the effect appears real but the evidence isn't settled. Speculative — cultural or design lineage only; no rigorous clinical evidence. We draw on it honestly, without dressing it up as science. The science page shows which techniques sit in which tier.

No. The “ancient solfeggio” frequencies are a 1970s reconstruction with no pre-modern lineage, and the 528 Hz “DNA repair” story isn't supported by mainstream science. We do use specific tone pairs in the headphone induction — because the binaural beat between two slightly-different tones has a real evidence base for procedural anxiety. Those are different things.

The clinical effects of binaural-beat audio on procedural anxiety are real and supported by meta-analysis. The EEG-entrainment mechanism sometimes invoked to explain those effects is contested in the literature. We don't conflate effect with mechanism — we describe what the audio does, and stop short of mechanism claims we can't defend.

iOS 17+ and Android 13+ at launch. The player uses platform-native media APIs for background playback, lock-screen controls, and AirPlay (iOS).

Compressed FLAC at launch, with MP3 as alternative. The canonical master is reserved for clean in-app playback on Monthly Practice and Companion plans.

No. Any reasonable pair that delivers stereo separation does the work for the induction phase. Better headphones give better imaging, but the composition is built to land on consumer-grade kit.

Sign in, open the Account hub, choose Delete your account. Account state, journal entries, and server-stored masters are gone within thirty days. Compressed downloads on your devices remain yours.

Yes. The Account hub has an Export tool that emails you a full bundle: account data, journal entries, recording metadata.

Use the “forgot password” link on the sign-in page; we send a reset link via Supabase. If nothing arrives within a few minutes, check spam, then email [email protected].