How accurate are phone decibel meters vs. dedicated SPL meters?
A dedicated Class 2 SPL meter is accurate to about ±1.5 dB by standard; good phone apps tested by NIOSH came within ±2 dB in lab conditions, but uncalibrated phone or browser readings can drift 5–10 dB in the real world. Phones are for awareness; calibrated meters are for evidence.
Last updated:
The accuracy ladder
From most to least accurate:
| Tool | Typical accuracy | Standard |
|---|---|---|
| Class 1 sound level meter | ±1 dB | IEC 61672 Class 1 |
| Class 2 sound level meter | ±1.5–2 dB | IEC 61672 Class 2 |
| Phone app with calibrated external mic | ±1–2 dB | NIOSH lab testing |
| Well-built phone app, internal mic | ±2–5 dB | NIOSH lab testing (best apps) |
| Browser-based meter (like ours) | ±10 dB | uncalibrated estimate |
The NIOSH lab numbers are the optimistic case: controlled environment, pink noise, phone held correctly. Out in the world — wind, reflections, a case over the mic, auto gain you can’t see — uncalibrated readings drift further.
What calibration actually does
A dedicated meter is calibrated end-to-end: its microphone capsule, preamp, and weighting filters are checked against a known reference tone (usually 94 dB at 1 kHz from an acoustic calibrator). A phone app can only apply a software offset on top of whatever the device’s microphone hardware and OS processing do — better than nothing, much better with a per-model profile, but never traceable the way a real meter is.
When each is the right tool
- Browser meter: instant answer to “roughly how loud is it here?” — try it now.
- Phone app: day-to-day exposure awareness, history, trends. Decibel Shield adds Apple Health sync and background measuring.
- Dedicated SPL meter: noise complaints with legal weight, occupational compliance (OSHA/NIOSH), product testing, anything someone might dispute.
Why our readings say “±10 dB”
We publish the honest number for an uncalibrated browser meter rather than implying instrument precision. Why apps disagree with each other covers where the variance comes from.