Online Decibel Meter — free sound level meter in your browser

A free online decibel meter with no ads and no signup

This online decibel meter measures sound level through your microphone, right in the browser. There is no account, no subscription, no free trial, no advertising and no app to install. It is a single web page that runs on a laptop, desktop or phone. Use it as a dB meter or noise meter for a room, a workshop, a rehearsal space, a classroom or a street outside your window.

Most free sound level meter pages show a big number without saying where it comes from. This one measures with a real IEC 61672 weighting filter, reports the metrics a real sound level meter reports, and is open about the one thing no web page can know on its own: how many decibels of real sound pressure a given digital signal means on your microphone.

How accurate is a browser decibel meter?

The frequency weighting is accurate. The A- and C-weighting curves are built as a cascade of biquad filters in the Web Audio graph and are normalised to 0 dB at 1 kHz. Compared with the standard curves they stay within 0.35 dB from 16 Hz to 4 kHz and within 0.85 dB at 8 kHz, and roll off about 3 to 4 dB early at 16 kHz, where everyday sounds carry very little energy.

The absolute level is only as good as your calibration. The sensitivity of your microphone depends on the capsule, the preamp gain and the driver, and none of that is visible to a web page. Until you calibrate, treat the numbers as a consistent relative scale: good for comparing one place with another on the same device. Once calibrated against a reference, repeat measurements on that device are consistent and useful. Do not use any browser meter for legal, medical or workplace-compliance decisions.

Why automatic gain control is switched off

By default a browser gives a web page a microphone stream tuned for voice calls, with automatic gain control, noise suppression and echo cancellation switched on. All three work against measurement. Automatic gain control raises the gain when a room is quiet and lowers it when the room is loud, which flattens exactly the variation you want to measure: a whisper and a shout can end up a few dB apart. This meter requests the microphone with autoGainControl, noiseSuppression and echoCancellation set to false, then reads back what the browser actually applied and shows it in the Controls panel. If your browser keeps one on, you see a warning instead of quietly wrong numbers. Chrome and Edge on desktop honour the request most reliably.

dBA, dBC and dBZ

These are frequency weightings. A-weighting turns down low and very high frequencies to match how sensitive human hearing is at moderate levels. Almost every noise regulation and hearing-protection guideline is written in dBA, and when an app or a rule simply says "decibels" it nearly always means dBA. C-weighting is much flatter and keeps the bass, so it is used for peak measurements and low-frequency problems such as club bass or plant rumble. Z is unweighted: the raw level with no filter.

LAeq, LAFmax and LCpeak explained

LAeq is the equivalent continuous level: the single steady A-weighted level that carries the same total sound energy as the whole varying measurement. It is the number noise surveys and regulations quote. It is not an average of the displayed readings: one loud event lifts LAeq far more than it lifts a simple mean.

LAFmax and LAFmin are the highest and lowest A-weighted levels seen with Fast time weighting. LCpeak is the true peak of the C-weighted waveform, taken at full audio sample rate instead of from the smoothed display. That is why it reads much higher than the level on screen for impulsive sounds such as a door slam or a hammer blow. The level work runs in an AudioWorklet that sees every sample, which is what makes LAeq and LCpeak trustworthy.

L10, L50 and L90

These are statistical percentiles of the measured level. L10 is the level exceeded 10% of the time and describes the intrusive peaks. L50 is the median. L90 is the level exceeded 90% of the time, the usual stand-in for the background once traffic, voices and other events are taken out. Environmental noise assessments rely on them, because a single average hides whether a place is steadily noisy or mostly quiet with the occasional bang.

How to calibrate the meter

Put this meter next to a reference you trust, such as a calibrated sound level meter or an acoustic calibrator producing 94 dB at 1 kHz. Expose both to the same steady sound, type the reference's reading into the calibration box and press match. The meter works out the offset and stores it in your browser for that device, so it is still there next visit. You can also type an offset directly, or press reset calibration to go back to the default.

Measuring with your phone

The page works in mobile browsers with microphone permission. Phone microphones are often better behaved than laptop ones, simply because they are not sitting next to a keyboard and a cooling fan. They do tend to limit and compress loud sounds, so a phone often under-reads concerts and power tools. Calibrate on the device you plan to use and treat readings above roughly 100 dB as approximate. If the input clips, the meter warns you to move further from the source.

For a big display, press F, click the reading or use the Fullscreen button. Where the browser refuses real fullscreen, as Safari on iPhone does, the readout fills the browser window instead.

How loud is too loud?

NIOSH recommends a limit of 85 dBA averaged over an eight-hour day, halving the allowed time for every 3 dBA above that: about 8 hours at 85 dBA, 1 hour at 94 dBA and 15 minutes at 100 dBA. The enforceable US OSHA standard is more permissive, with 90 dBA over eight hours and a 5 dB exchange rate. For homes rather than workplaces, the WHO 2018 European guidelines recommend keeping road traffic noise below 53 dB L_den and 45 dB L_night. The reference table on this page lists typical sources and NIOSH exposure times from 10 to 140 dB.

Export your measurement to CSV

Export CSV downloads a file with a header describing the session (weighting, time weighting, calibration offset, duration, LAeq, Lmax, Lmin, LCpeak and L10, L50, L90) followed by one row per second with the mean level, the maximum, the running LAeq and the running LCpeak. Open it in a spreadsheet to chart a longer measurement or attach it to a report.

Private: audio never leaves your device

The microphone stream is processed by the Web Audio API inside your browser tab. Nothing is written to disk or sent anywhere, and there is no server to send it to. Closing the tab discards everything. The only things stored locally are your calibration offset and display preferences. The page's only analytics is SimpleAnalytics, which uses no cookies.

Keyboard shortcuts

Space start or stop measuring · R reset the session · E export CSV · F fullscreen · Esc leave fullscreen.

More audio tools

Need a quieter room to work in? Try Focus Audio for background noise and focus sounds. Musicians can tune up with the online guitar and instrument tuner and keep time with the metronome and BPM counter. Watch sound move in the music visualizer or cut a recording in the audio editor. See all music tools or the rest of the collection.