How we record and measure

Two kinds of work happen behind Sonora: making the sound, and measuring how people respond to it. Both deserve to be documented.

Recording

Every soundscape starts as a recording of a real place, captured with a Soundfield ambisonic microphone — a four-capsule design that records the full three-dimensional sound field, not a stereo slice of it. That is what lets a recording keep its sense of space on headphones.

Takes are long-form by design. A loop a listener can spot is a loop that wakes them, so recordings run long enough — and are edited carefully enough — that no repeat point is there to find. Nothing in the library is synthesised and nothing is generated by AI; when you hear rain, it fell somewhere.

Measurement

The research side works with replay data: which sounds people combine, at what levels, and whether they keep coming back over months. For the Forum Acusticum 2026 study, that meant twelve months of data from 77,408 users, characterised acoustically (level, spectral balance, and the standard ISO 532-1 psychoacoustic descriptors) and related to return behaviour on held-out data.

Replay behaviour is a proxy for preference, not a sleep measurement — the honest limit of this method, and stated on every study page.

Privacy

Listening data is anonymised before analysis and studied in aggregate. The research asks how people in general respond to sound; it has no use for who you are, and the microphone on your device plays no part in it.