How loud is it — really? A pocket noise-level logger that measures how loud a space is using your device's microphone and charts the level over time — with the minimum, average and peak for each session. 100% client-side, zero dependencies, works fully offline. Nothing is recorded or uploaded.
You can feel that a room is loud, but you can't compare it — the café you thought was quieter than the office, the nursery at 9pm versus 2am, the fan on low versus high. Most phone noise apps bury a simple reading under ads, accounts, and permissions you can't verify, and many quietly upload data.
hushmeter is the opposite: open the page, allow the mic, and watch a live level with a moving waveform. Press Start session to log the level over time and get a clean min / average / peak, then name it ("Baby's room, 9pm") and save it. Everything — the audio analysis and the saved sessions — happens on your device. You can prove it: a strict Content-Security-Policy forbids the app from making any network request.
- Live meter — a big current level, a moving oscilloscope-style waveform, and a colour-plus-text band (Quiet / Moderate / Loud / Hearing risk) so it never relies on colour alone.
- Sessions over time — start and stop a measurement; hushmeter samples the level a few times a second and draws a timeline chart, tracking running min, average and peak.
- Saved history — name and tag each session, reopen it to view its chart, delete it, or export it as CSV (
timestamp,level) or a text summary. - Calibration offset — a simple dB offset slider so, if you have a reading you trust, you can align hushmeter to it. The offset is remembered on your device.
- Reference guide — an indicative table of what typical dB levels mean, from a whisper (~30) to hearing-risk territory (85+).
- Private by construction — no accounts, no network, no analytics, no external assets. The microphone stream is connected only to a Web Audio analyser — never to a media element — so nothing is ever played back, recorded, or written to a file.
Just open index.html in any modern browser — no build step, no server, no install. Grant microphone permission when asked.
- Local: double-click
index.html, or run a static server in the folder (some browsers only allow mic access overhttp://localhostorhttps://, so a tiny local server is the most reliable way to try it). - Hosted: Open hushmeter live
Your saved sessions and calibration offset live in your browser's local storage on this device only. They are not synced or backed up — export a session to CSV if you want a copy to keep.
hushmeter is built so you don't have to take its word for it.
- A strict Content-Security-Policy sets
connect-src 'none': the app cannot make any network request, even if it tried. - The microphone stream is routed into a Web Audio
AnalyserNodeonly. It is never attached to an<audio>or<video>element, never sent to the speakers, and never recorded to a file or aMediaRecorder. - No external fonts, scripts, images, or analytics. Everything is self-contained and runs offline.
- Only the numeric session summaries you choose to save are stored, and they stay in local storage on your device.
hushmeter is not a calibrated sound-level meter. Consumer microphones are uncalibrated, their frequency response varies wildly, and most phones and laptops apply automatic gain control that raises quiet sounds and pulls down loud ones. hushmeter turns the microphone's signal into a relative loudness indicator and even applies a modest smoothing and an optional calibration offset — but it can never be certified decibels.
Use it to compare: this room versus that one, now versus later, before versus after you change something. Do not use it to decide whether a workplace is within legal noise limits, whether a sound is medically harmful, or anything else where a real measurement matters. For that, use a calibrated SPL meter and follow the applicable regulations.
hushmeter provides a relative, indicative loudness reading for general awareness and educational purposes only. It is not a calibrated sound-level meter and is not professional acoustic, occupational-health, medical, engineering, or legal advice. Readings are uncalibrated and affected by your microphone and automatic gain control, and must not be used for occupational, medical, or legal noise-compliance decisions. This software is provided under the MIT License, "as is", without warranty of any kind; the authors accept no liability for any loss, injury, or damage arising from its use.
MIT © 2026 Sreenivas Sadhu Prabhakara