How to Calibrate a Decibel Meter (Step-by-Step, Including Phone Apps)

What Calibrating a Decibel Meter Actually Means

A decibel meter doesn't measure sound pressure directly. It measures the voltage its microphone produces and converts that to a level using the microphone's sensitivity — how many millivolts one pascal of sound pressure produces. That figure is what calibration establishes, and it moves: with temperature and humidity, as the diaphragm ages, and after a knock. Calibrating means exposing the microphone to a known sound pressure and correcting the reading to match.

Three different jobs go by the same name, and most confusion about calibration comes from mixing them up:

What You Need to Calibrate a Sound Level Meter

• The meter's manual. It gives the level to set, the settings to calibrate on, and how to make the adjustment. These differ between models, and where the manual disagrees with the steps below, follow the manual.
• An acoustic calibrator. A hand-held device that fits over the microphone and produces a fixed tone at a known level inside a sealed cavity. Calibrators are specified by IEC 60942 (ANSI/ASA S1.40 in the US) in classes that mirror the meters': use a Class 1 calibrator with a Class 1 meter; a Class 2 meter can use either.
• The right adapter. A calibrator is built around one microphone diameter, commonly 1/2 inch, and takes adapters for others. Use the one that matches your meter's microphone — a loose fit leaks and reads low.
• Somewhere to write the results. A notebook or a spreadsheet. The record matters as much as the check.

Method 1: How to Calibrate a Sound Level Meter With an Acoustic Calibrator

This is the check every sound level meter manual describes, and the one to do at the start and end of every set of measurements. It takes about a minute.

1. 1

   Let the equipment acclimatise.

   Bring the meter and calibrator into the space where you will measure and leave them for several minutes — longer if they have come in from a cold vehicle. Condensation on a microphone diaphragm changes its sensitivity, and a calibrator's output shifts slightly with temperature and air pressure.

2. 2

   Check the batteries in both.

   Replace anything low. A calibrator on a weak battery may not hold its tone, and a meter that shuts down halfway through a session loses the session.

3. 3

   Set up the meter.

   Remove the windscreen. Select the settings the manual gives for calibration: typically A or C weighting, Fast or Slow, and a measurement range that includes the calibrator level. Many current meters have a calibration mode that sets these for you.

4. 4

   Fit the calibrator.

   Push it straight and gently onto the microphone until it seats fully, with the adapter if your microphone needs one. The seal is what matters, not force. Rest the meter on a table, or hold it by the body rather than around the microphone end.

5. 5

   Switch on and choose the level.

   Select 94 dB, or 114 dB if the surroundings are loud (see below). Wait a few seconds for the tone and the reading to settle.

6. 6

   Compare, and adjust if needed.

   The target is the level on the calibrator's calibration certificate, adjusted by any correction the meter's manual gives for its microphone — which is why the figure to set is sometimes a tenth or two away from 94.0. If the reading is outside the manual's tolerance, adjust the meter until it matches: through a menu on most current meters, or a small trim screw on older ones.

7. 7

   Note the result, then measure.

   Record what the meter read before and after any adjustment. Remove the calibrator, refit the windscreen, and take your measurements.

8. 8

   Check again afterwards, without adjusting.

   At the end of the session, fit the calibrator again and record what the meter reads. Don't correct it. The difference between the before and after readings is how far the meter drifted during the session, and it is the evidence that the measurements in between are sound. Many measurement procedures treat a difference of more than about 0.5 dB as grounds to repeat the measurements; check the one you are working to.

9. 9

   Keep the record.

   Date and time, the meter's and calibrator's serial numbers, the pre- and post-check readings, and any adjustment made. For workplace or legal measurements, this log and the laboratory certificates are what make the numbers defensible.

94 dB or 114 dB?

94 dB is a sound pressure of 1 pascal. It is the standard reference level and the one most manuals specify. 114 dB is 10 pascals, 20 dB higher. Use it where background noise — a production line, a generator, traffic — is loud enough to reach the microphone past the calibrator and nudge the reading. Either is valid if the meter's range covers it. What matters is using the same level for the check before and the check after.

Why 1 kHz?

A-, C- and Z-weighting all pass 1 kHz unchanged, so a 1 kHz tone reads the same whichever weighting is selected, and the check doesn't depend on a setting people often get wrong. That isn't true of pistonphones, a mechanical type of calibrator that usually runs at 250 Hz. A-weighting takes about 8.6 dB off at that frequency, so a pistonphone is read with C or Z weighting, and its output also needs correcting for barometric pressure as its documentation describes.

Field Check vs Laboratory Calibration

A calibrator check tests one level at one frequency. It confirms that the microphone and the meter's gain are where they should be, which is the part most likely to drift. It says nothing about whether the frequency weighting is still right across the audible range, whether the meter is linear from its quietest range to its loudest, or whether the time weighting responds as it should. Nor can it tell you whether the calibrator is still producing the level on its label.

That is what laboratory calibration covers. The meter goes to a calibration laboratory, usually with its calibrator, and is tested against reference equipment traceable to a national standards body — NIST in the US, NPL in the UK. The tests for sound level meters are set out in IEC 61672-3. Choose a laboratory accredited to ISO/IEC 17025 if the certificate may be scrutinised, and expect a few hundred dollars.

Most manufacturers and measurement methods ask for a laboratory calibration every one to two years, and commonly every year for the calibrator, since every field check depends on it. Send the meter sooner after a drop, a repair, a failed field check, or a run of pre- and post-checks that drift in the same direction.

For workplace noise in the US, OSHA's occupational noise standard, 29 CFR 1910.95(d)(2)(ii), requires instruments used to measure employee noise exposure to be “calibrated to ensure measurement accuracy.” It doesn't set an interval or a method, so in practice the manufacturer's instructions, the field-check log and the laboratory certificates are how an employer shows it was done. The rest of what the standard requires is in our guide to OSHA noise limits.

How to Calibrate a Phone or Browser Decibel Meter

A phone's microphone sits behind a small opening in the case, not in a capsule a calibrator can fit over. A calibrator's level depends on a sealed cavity of known size, and held against a phone it leaks by an unknown amount. Phones, apps and browser meters are calibrated by comparison instead: measure the same sound with a calibrated meter and your device, in the same place at the same time, and the difference is your device's offset.

Our online decibel meter has no calibration setting. It converts the microphone signal to a sound level by assuming that a full-scale signal equals 94 dB SPL, the same fixed reference for every device — the methodology page explains why. So the offset you find is one you apply by hand: write it down and add or subtract it when you read the meter. Some phone apps have an offset or calibration field; if yours does, enter it there.

1. 1

   Get a reference.

   A Class 1 or Class 2 sound level meter, checked with its calibrator the same day. Test-equipment rental companies hire them out. If a calibrated meter isn't available, a recent iPhone running a well-implemented measurement app, such as NIOSH's free SLM app, is a weaker but usable stand-in; the published testing behind that is summarised on our accuracy page.

2. 2

   Choose a steady, broadband sound.

   Pink or white noise from a speaker is ideal; a fan, a cooker hood or a running shower will do. Avoid speech and music, which change too quickly for two meters to agree on, and avoid a single tone: in a room, a tone sets up standing waves, so its level changes over a few centimetres.

3. 3

   Set a sensible level.

   Somewhere between about 60 and 85 dBA at the measuring position. That is well above a phone microphone's noise floor, and well below the roughly 100 to 105 dB where phone microphones start to compress.

4. 4

   Put both microphones in the same place.

   Side by side, a few centimetres apart, pointing the same way, at least a metre (3 feet) from the source and away from walls. Find your phone's microphone first — usually on the bottom edge — and make sure neither a case nor your hand covers it. A stand or a stack of books is better than holding either device.

5. 5

   Match the settings.

   Set the reference meter to A-weighting, since this meter reads dBA. Comparing dBA on one with dBC on the other is a common reason for a comparison to go wrong. Time weighting matters less with a steady sound, though Slow makes the reference easier to read.

6. 6

   Measure for at least 30 seconds and compare averages.

   Let both run, then compare this meter's Average with the reference meter's Leq or average. With a steady sound the two agree to within a fraction of a decibel, whichever kind of average each one uses.

7. 7

   Work out the offset.

   Offset = reference reading − phone reading. If the reference shows 72.4 dBA and the phone 69.1 dBA, the offset is +3.3 dB: add 3.3 dB to that phone's readings.

8. 8

   Repeat at two or three levels.

   Turn the source up and down within the range above. If the offset holds to within about a decibel, one number will do. If it grows at the loud end, the microphone is compressing; if it grows at the quiet end, you are near its noise floor. Either way, trust the offset only across the range where it held.

9. 9

   Write it down.

   The offset, the device model, the operating system version, the browser or app, and the date. The offset belongs to that combination: another phone, another browser or an operating system update can change the gain before the meter sees the signal, and a new case can cover the microphone differently. Check again after any of them.

Why one offset doesn't fix everything

A single offset corrects the phone's sensitivity: how much signal it produces for a given sound pressure. It does not correct its frequency response. Phone microphones are built for speech. They are typically less sensitive at low frequencies and uneven at high ones, and the pattern differs between models. The offset you measured is right for sounds with a spectrum like your calibration source's, and less right the further a sound departs from it — the low hum of a ventilation plant, or the high whine of a machine.

That is why broadband noise makes a better calibration source than a tone: it spreads its energy across the band much as most real noise does. A-weighting helps too, since it already discounts the low frequencies where phone microphones are weakest. What an offset can't do is raise the ceiling. A phone microphone that compresses above 100 dB still does, whatever you add to the reading.

An offset also only works if the gain stays fixed. Automatic gain control, which browsers and many apps switch on by default, turns the gain up in a quiet room and down in a loud one, so the offset changes with the level. Our meter switches off automatic gain control, noise suppression and echo cancellation for that reason. If an app's offset moves by several decibels between levels, gain control is the likely cause. The accuracy page sets out what a calibrated browser reading can and can't be relied on for.

No Reference Available? A Sanity Check, Not a Calibration

Lists of familiar sounds are often offered as a way to calibrate without equipment. They can't do that job, because real sources vary too much: one vacuum cleaner can be 10 dB louder than the next. What they can do is catch a device that is badly wrong.

If a device reads 15 or 20 dB away from these in several places, something is wrong: a covered microphone, a headset or external input selected instead of the built-in mic, or gain control left on. If it reads inside the ranges, that tells you little more, because the ranges are wider than the error you are trying to find. For typical levels of other everyday sounds, see the decibel chart.

Common Calibration Mistakes

How Often Should You Calibrate?

• Sound level meter, field check: before and after every measurement session.
• Sound level meter, laboratory calibration: at the interval the manufacturer or your measurement method sets, usually every one to two years, and after any drop, repair or failed field check.
• Acoustic calibrator: laboratory calibration, commonly every year.
• Phone or browser meter: once per device and browser, then again after an operating system update, a browser change or a new case, or whenever your own reference sound starts reading differently.

If you are still deciding whether you need a hardware meter at all, our comparison of online and hardware decibel meters sets out where each one fits, and the best sound level meters guide covers the Class 1 and Class 2 instruments worth buying.

Frequently Asked Questions