Measure noise with your microphone: live dB, plus minimum, maximum and average
Processed in your browser. Nothing is uploaded. Updated on 24 Oct 2026.Measuring noise properly needs a calibrated reference, which a phone microphone does not have, so read the number as indicative. Live decibels are shown with the minimum, maximum and average. Above 85 dB across a working day is where hearing damage begins.
Press Start the meter and allow the microphone when your browser asks.
Read the live level in dB(A) and the bar under it; the line beside it compares the level with everyday sounds.
Leave it running to fill in Minimum, Maximum and Average, and watch the Last minute strip.
Press Stop and release the microphone when done; the sound is measured on your device and nothing is recorded or sent.
This sound level meter reads the microphone your device already has and shows a live level in decibels, with the quietest moment, the loudest moment, an energy average and a strip showing the last minute. Everything runs in the page: no audio is recorded, stored or sent anywhere, and the microphone is released the moment you press stop, so the browser's recording indicator goes out.
A phone or laptop microphone has no calibration certificate, no published sensitivity, and an automatic gain control the operating system can move without telling the browser. The absolute figure therefore rests on one assumption: that a full scale sample is 100 dB SPL. That assumption puts a quiet room near 35 dB and a conversation near 60 dB on most devices, which is about right, but it is an estimate and not evidence. A landlord, an employer or a court needs a calibrated class 1 or class 2 meter, not a browser tab.
What is reliable is the comparison. Two readings taken on the same device within a few minutes share the same unknown sensitivity, so the difference between them is real. That is enough to answer most of the questions people actually have: is the bedroom quieter with the window shut, is this cafe noisier than the last one, did moving the desk away from the vent help at all.
A-weighting throws away bass the ear barely registers and lifts the 2 to 4 kHz region the ear is most sensitive to, which is what almost every published noise figure uses. Switch to Z-weighting to see the level with nothing removed, which is the honest choice when you are chasing a low-frequency hum. Fast response follows sound over 125 milliseconds and slow over a second; slow is easier to read and is what noise surveys use.
If you own a real meter, put the two side by side and use the calibration offset to line the scales up. Everything afterwards is then measured against something you trust. Related tools: Room Echo, Tone Generator.
Accuracy in absolute terms is poor, because consumer microphones are uncalibrated and their sensitivity is unpublished. Expect a reading to sit within roughly ten decibels of a real meter. Comparisons taken on the same device minutes apart are far better, because both readings carry the same unknown offset and it cancels out.
No. Noise limits in tenancy agreements, workplace rules and local law are written against calibrated class 1 or class 2 meters, and a browser reading would not be accepted anywhere. Use the meter to decide whether a complaint is worth making, then ask the council or your employer for a proper measurement.
A-weighting is a curve that reduces the contribution of low frequencies and slightly lifts the 2 to 4 kHz region, matching how sensitive human hearing is at each frequency. Almost every published noise figure, from traffic limits to workplace exposure, is A-weighted, so use it when comparing with anything you have read.
Averaging happens in energy, not in decibels, which is what a real meter reports as the equivalent continuous level. One loud moment carries far more energy than several quiet ones, so the average sits closer to the loud end. An arithmetic mean of decibels would understate a noisy room.
Nothing at all. Sound is measured frame by frame inside the page and discarded immediately, no audio buffer is kept, no file is written, and no request carries audio anywhere. Pressing stop releases the microphone track, which is what makes the browser's recording indicator switch off.
Quiet bedrooms measure about 25 to 35 dB, a library around 40, a quiet office around 50, and normal conversation about 60 decibels a metre away. Busy traffic from the pavement reaches roughly 70 to 80. The United States National Institute for Occupational Safety and Health sets its exposure limit at 85 dB averaged over eight hours.
Still stuck? Ask about Sound Level Meter and anybody who knows can answer. No account needed.
Also works as: Decibel Meter · dB Meter Online · Noise Level Meter.
Same job, other tools: Hz Generator – Type a frequency in hertz and hear it, from 20 Hz to 20,000 Hz.
On the related tools: Speed Up Audio – Make audio faster or slower without the chipmunk voice · Bass Booster – Add more low end to a song or recording, with the clipping kept in check, in your browser.
In the collection Audio Test Tools.