Snoaring starts recording when you fall asleep, keeps only the snoring it hears, and has a report waiting when you wake up.
Set your sleep window once. When your phone goes still at night, recording begins — no button at bedtime.
A small neural network analyses the audio locally. Continuous audio is never uploaded, and no connection is needed to record a night.
Instead of storing the whole night, it saves short clips of the episodes it detects, so you can listen back to what actually happened.
Episode count, time spent snoring, loudness and a snore index you can track night to night.
Loudness alone cannot tell a snore from a door closing or a passing car, so Snoaring uses two stages. The first watches for loud bursts against the room's own noise floor and checks whether they arrive with the rhythm of breathing. The second asks a neural network, trained on thousands of real recordings, whether each burst actually sounds like snoring. Only bursts that pass both, and that repeat in a breathing pattern, become an episode.
Both stages run on your phone. The model is small enough to sit inside the app and cost almost nothing to run all night.
Snore clips and reports are stored on your device. When a nightly report is generated, a few of the loudest clips are sent to Google's Gemini so it can describe your snoring and suggest changes — that is the only audio that ever leaves your phone. The full detail is in the privacy policy.