Monitoring noise without recording conversations

A host operating short-term rentals kept receiving noise complaints from neighbors, but had no way to know a problem was happening until someone else reported it — often after the fact.

Short-term rental properties sit inside a web of constraints that aren't purely technical: guest privacy expectations, platform policies on surveillance, and local rules that vary by jurisdiction. Guests also need to be told, clearly, what's present in a space.

The instinct is to reach for a microphone that records. That's the wrong tool here: it solves the problem technically while creating a privacy and policy problem that's worse than the noise itself. The actual requirement is narrower — know when sound levels cross a threshold, not know what was said.

A small edge device with a sound-level sensor analyzes decibel levels locally and discards the raw audio immediately — nothing conversational is ever stored or transmitted. When levels cross a sustained threshold, the device sends a lightweight event (not audio) to a simple dashboard, so the host can follow up in real time instead of after a complaint.

• No recording or storage of conversational audio, under any circumstance
• Guests must be clearly disclosed to before check-in, not after
• Hosting platform terms restrict certain forms of in-unit monitoring
• Has to run unattended for weeks between cleanings, on battery or low power
• False alerts have a real cost — they erode trust in the system fast

A first pass would run on a single unit in one property: confirm the sensor's threshold tuning matches what actually counts as 'too loud' in that space, and confirm alerts arrive fast enough to be useful before scaling to more units.