How to Calibrate a Phone Decibel Meter

Calibrating a phone decibel meter means one thing and one thing only: finding out how far your device is from something you trust, and carrying that difference forward as a fixed offset.

It does not mean making the reading correct. It cannot, and a page promising otherwise is selling something. What it does is remove the constant part of the error, which on most devices is the largest part — and that is enough to change what the number is good for.

What calibration can fix, and what it cannot

The error in a phone reading is not one thing. Some of it is a constant offset — your microphone reads consistently a few decibels high or low across the whole range. That part is fixable, and it is what an offset removes.

The rest is not. A microphone's frequency response is shaped for speech and nobody publishes the shape, so the error changes with the kind of sound as well as its level. A single number cannot correct a curve. Automatic gain control, when a device refuses to switch it off, compresses loud passages by a varying amount. And once the input runs out of headroom, no offset in the world brings back a level the hardware never captured.

So the honest framing is: an offset moves the whole scale, and everything that is not a constant survives it. That is still worth doing, because after it the comparisons you make on one device get their scale roughly right rather than merely consistently wrong.

The evidence that offsets matter comes from the study everyone in this field cites. When Kardous and Shaw selected apps to compare against a reference, they specifically included ones that allow calibration adjustment of the built-in microphone through manual input or uploaded filesKardous & Shaw. It is the standard feature of a serious sound app for a reason.

The method, without lab equipment

You need one thing: a sound whose level you already trust, holding steady long enough to read.

1. Find your reference. In descending order of how much you should trust it: an acoustic calibrator that fits your microphone, a sound level meter you already own, a borrowed meter, or a meter belonging to someone whose job involves one. Each of these produces a number you can stand next to.
2. Put both in the same place. Same spot, same height, same direction, at the same time. A metre of distance between the two microphones can be worth several decibels on its own, which would go straight into your offset as a permanent error.
3. Use a steady sound. Not music, not a room, not traffic. Something that holds a level: a calibrator, a fan, a tone. A fluctuating source means you are comparing two different moments and calling the difference a device error.
4. Read both for at least thirty seconds. Let the numbers settle and take where each one sits, not what either touches.
5. Take the difference and set it as the offset. If the reference reads 82 and the phone reads 76, the phone is six decibels low and the offset is +6. On this site that goes in the calibration offset control under the meter, and it is stored on your device only.

If you own an acoustic calibrator, steps one to four collapse into fitting it to the microphone and reading the number: a calibrator exists precisely to produce a known level at a known frequency. What this page will not do is tell you what level yours produces — that is printed on the device, and inventing a figure here would be exactly the kind of unsourced number the rest of this site refuses to publish.

Calibrating against a phone app is not calibrating

A common shortcut is to open a well-reviewed app on another phone and match to that. It feels like a reference. It is not.

The study that compared ten iOS apps against a genuine type 1 sound level meter found the best of them sitting about half a decibel from the referenceKardous & Shaw — but only three of the ten came within two decibels, and the rest were scattered well beyond thatKardous & Shaw. Picking one at random and treating it as truth means inheriting whatever error it has, on top of your own, with no way to tell which is which.

If the other phone has itself been calibrated against something real, it is a reference. If it has not, you are averaging two unknowns.

How to tell whether it worked

Check it against a second point at a different level.

An offset is a single number applied across the whole scale, so if it is right in a quiet room it should still be roughly right in a loud one. If the two points disagree — the phone matches the reference at conversation level and is several decibels out beside a machine — the error is not a constant, an offset will not fix it, and you have learned something more useful than a number: your device's error changes with level, and readings at the loud end deserve less trust than readings at the quiet end.

Two other signs are worth knowing. If the meter warns that automatic gain control could not be switched off, calibrate anyway but expect the offset to hold only for steady sounds. And if the reading stops responding as things get louder, the input has clipped — recalibrating changes nothing, because the number was never about the room at that point.

What calibration does not buy you

It does not make the reading admissible. NIOSH's own position is that where a sound level meter is not available, sound measurement apps can provide a measure of area noise but may not comply with regulatory requirementsNIOSH. A calibrated phone is still a phone; the sentence does not stop applying because you subtracted six decibels.

It also does not close the gap to instrument hardware. A Class 2 sound level meter is built around a microphone with a published response, it holds calibration, and it is designed to be checked. If the number has to survive being challenged by someone with an interest in challenging it, that is what the number has to come from.

What calibration does buy is a scale you can reason about. After it, a reading of 85 on your device means something closer to 85 in the room, and the band around it is narrower than the ±10 dB an uncalibrated estimate carries. The full argument for that band is under how accurate is an online decibel meter.

Recalibrating, and how often

Rarely, and for specific reasons rather than on a schedule.

A phone's microphone does not drift the way a precision instrument does, because it was never that stable to begin with — but the things around it change. A new case, a different phone, an operating system update that changes the audio path, or a blocked microphone port are all reasons to check again. So is any reading that surprises you in a room you know.

The habit worth building is not frequent recalibration. It is keeping the offset attached to the device it was derived on, and not carrying a number from one phone to another. The offset in the browser decibel meter on this site lives in that browser on that device for exactly this reason: it is a fact about your hardware, not about the site.