How to Calibrate a Phone Decibel Meter (Reference-Only)

The 30-second answer: one offset, one reference

Every phone microphone has its own sensitivity, so two phones on the same table commonly disagree by 5–10 dB. Calibration corrects that systematic bias with one stored number: measure a sound whose level you trust, subtract your reading from the reference, and add that offset to every future reading. Done once per device and setup, it turns "louder vs quieter" into a usable absolute estimate — typically within about ±2 dB of a professional meter on steady mid-range sounds.

What it is not: certification. Calibration fixes overall sensitivity at the reference level only. It cannot flatten bass roll-off, extend the capsule's dynamic range, undo automatic gain or noise suppression, or qualify numbers for courts, regulators, or workplace exposure. Treat calibrated phone data as screening data with an error band — full limits in our accuracy guide and calibration guide.

What counts as a trusted reference

You need a sound whose level is known independently of the phone you are calibrating. Ranked from best to weakest:

Whatever you use, keep the phone microphone exactly where the reference level is valid, keep the sound steady for at least 10 seconds (30–60 is better), and quote the average or Leq — never a flickering instant. See why readings jump for Fast vs Slow vs Leq.

The offset math: one subtraction, one check

Offset (dB) = Reference level (dB) − Your meter reading (dB). Example: the calibrator outputs 94 dB and your phone reads 88 dB → offset = +6 dB. The meter adds 6 dB to every reading from now on. If your phone read 97 dB against the same 94 dB tone, the offset is −3 dB.

Then verify: re-measure the same reference without moving anything — the meter should now agree within about 1 dB. If it does not, something moved, the sound was unsteady, or processing interfered (see below). Record the device, case, OS version, reference, offset, date, and position — that line is what makes next month's readings comparable. Log repeats in the Noise Journal with CSV or JSON exports so outliers stay visible.

Step-by-step routine in this tool

1. Set up. Uncover the mic pinhole (remove bulky cases or align cutouts), hold the phone at ear height where people actually are, silence yourself, keep the tab foreground with the screen awake, and pick dBA for everyday noise or dBC when bass dominates (see dBA vs dBC). 2. Open Calibrate in the meter card header. 3. Enter the reference level (94 dB is pre-filled). 4. While the reference plays, note the current reading and enter it. 5. Apply Calibration — the offset is computed and stored on your device only. 6. Re-measure the reference to confirm.

Phone-specific notes: iPhones are the most consistent because few models share similar capsules — per-model presets transfer reasonably. Android spans hundreds of capsules, so calibrate each device individually and never copy offsets between phones. Browser meters request echoCancellation: false, noiseSuppression: false and autoGainControl: false, but some phones or browsers ignore the request — check the Device and Microphone panel, and if the reading never moves or pumps with speech, suspect frozen gain. Full workflow in how to use an online meter and how to measure decibels.

Make it repeatable: the 6-point checklist

1. Same hardware, same case, same orientation. A different case or a rotated phone is a different instrument — recalibrate. 2. Documented distance. Write "63 dBA Leq 60 s at 1 m, mic toward source" every time. 3. Average, don't snapshot. Quote Leq over 30–60 seconds plus minimum for background and maximum for events. 4. Two spots. Repeat from a second position; agreement within a few dB means the picture is trustworthy. 5. Verify each session. A 10-second re-check against a stable sound catches drift after OS updates. 6. Sanity-check against known levels in everyday sounds and the full chart — a bedroom reading 70 dB means a blocked mic or nearby fan, not a loud bedroom.

Placement dominates real-world error more than the app: pointing away, pocketing the phone, grip handling noise, wind outdoors, and hard-wall reflections each add several dB. Fix placement before blaming calibration — details in are apps accurate and how we measure.

What calibration cannot fix — and when to call a pro

Calibration corrects sensitivity, not physics: uneven frequency response (phones under-read bass), limited range (unreliable below ~30 dB and above ~95–100 dB), hidden processing that shifts gain mid-session, and room or wind artefacts all remain. Recalibrate after phone, OS, browser, or case changes — and monthly for any series you compare.

Never use phone numbers — calibrated or not — for workplace exposure, planning or nuisance disputes, product specs, commissioning, or anything that will be challenged. Those need an IEC 61672 meter with laboratory calibration and a competent surveyor reporting Leq plus L90, L50, and L10 with uncertainty. Use your phone log as a dated briefing ("living room 63 dBA Leq 15 min, peaks 78 dBA, window closed, 22:10–22:25") so the professional knows exactly where to deploy certified kit. See how we validate for our own chain and limits.

Reference tool only — not for legal, medical, occupational, or regulatory decisions.

Frequently asked questions

Sources and further reading

• IEC 61672-1 — Electroacoustics: sound level meters; A-weighting and Leq definitions used for reference levels.
• Kardous and Shaw 2014, JASA — smartphone app evaluation (4 iOS apps within 2 dB on pink noise 65–95 dB).
• Kardous and Shaw 2016, NIOSH follow-up — NIOSH SLM within 1 dB with calibrated external microphone.
• NIOSH — Criteria for a Recommended Standard: Occupational Noise Exposure (85 dBA limit, 3 dB exchange rate).