Accuracy & calibration

The online meter is an estimate, not an instrument. Browser and phone microphones are uncalibrated, so expect readings within roughly ±10 dB of a true sound level meter, with the largest errors at very quiet and very loud extremes. That is plenty for comparing situations, but not evidence for a legal or workplace noise complaint.

Last updated: June 13, 2026

This page is the honest, detailed version of the disclaimer under the meter. For how the reading is computed end to end, see the methodology; for our sourcing rules, the editorial policy.

How accurate is the online decibel meter?

Treat readings as accurate to within about ±10 dB of a calibrated sound level meter. The error is not uniform: it is smallest in the mid-range (roughly 50–85 dB, where most everyday noise sits) and largest at the extremes — very quiet rooms, where the microphone’s own noise floor dominates, and very loud sounds, where the input clips. That band is good enough to answer “is this room about 60 dB or about 90 dB?” It is not good enough to assert “this is exactly 73 dB,” which is why the readout rounds and the disclaimer never leaves the screen.

Why aren’t browser microphones calibrated?

Consumer microphones are built for voice, not metrology. Their frequency response is tuned for speech, their sensitivity varies from device to device, and browsers normally apply automatic gain control and noise suppression that move the level around. We disable that processing where the browser allows it, so the reading tracks the real signal — but there is still no calibrated reference inside a phone or laptop. A true sound level meter is calibrated against a known reference tone (a calibrator) to meet the IEC 61672 standard; a browser tab cannot do that.

How does Decibel Shield calibrate the reading?

The reading starts from the microphone signal’s energy in dBFS and adds a calibration offset to approximate dB SPL. That offset has three parts:

A base offset tuned so a typical device reads in the right ballpark.
A per-device default. iPhone and iPad microphones report roughly 11 dB lower digital levels than laptop microphones for the same sound (measured against a reference meter, 2026-06-09), so we add a platform offset on iOS automatically.
Your manual trim. The +/− control under the meter nudges the reading up or down (up to ±20 dB) to match a reference meter you trust. It is saved on your device and survives reloads.

None of this makes the meter a calibrated instrument — it makes the estimate land closer to reality on your hardware. Why two phone apps disagree by 5–10 dB on the same sound comes down to exactly these factors; the FAQ covers why decibel apps show different readings.

What is the meter good for — and not good for?

Good for: comparing one environment to another (is the café louder than the street?), spotting when a space is in the hearing-risk range, ballpark readings for curiosity, classrooms and personal awareness. The live NIOSH verdict and the decibel levels chart put any reading in context.

Not valid for: legal evidence in a noise dispute, OSHA or NIOSH occupational compliance, or any contractual measurement. Those require a calibrated meter meeting IEC 61672 (Class 1 or Class 2). An uncalibrated browser reading will not hold up, and we will not pretend otherwise.

How do I get a calibrated reading?

For documentation that needs to stand up, use a dedicated SPL meter, or a meter app whose calibration you have checked against one. The Decibel Shield iOS app adds adjustable calibration, measurement history and exposure alerts on a platform where the microphone behaviour is more consistent than the open web. For how a phone compares to a dedicated meter in general, see phone vs dedicated SPL meter.