Questions to ask before buying a sound meter or a noise survey

Before buying anything that measures sound, answer these. Several of them will send you back to the free option, which is usually the correct outcome.

What to take away

Most sound measurement apps are not accurate enough for anything official, and a small number are.
An external calibrated microphone changes the answer more than a different app does.
Ask who must accept the result before choosing an instrument.
If the answer will only ever be compared to itself, buy nothing.

Question one: who has to believe this?

If the answer is you and one other person in the house, a phone is enough. If the answer includes a landlord, an employer or an insurer, the instrument needs a calibration record and the method needs to be written down. That single question decides most of the budget.

Question two: has this app been tested against a reference?

Very few have. A study that set out to test this systematically acquired a representative sample of phones and tablets, screened more than 130 iOS applications and found only ten that met its selection criteria, with only four of 62 Android applications tested at all. Of those that were tested, two showed mean differences of 0.07 dB unweighted and minus 0.52 dB A-weighted against the reference system, and two more were within plus or minus 2 dB. The conclusion was narrow and useful: certain apps may be appropriate for occupational measurements, and most are not.

Question three: microphone or app?

If you are going to spend anything, spend it here. A follow-up study took the four apps that had performed best and re-tested them using external calibrated microphones rather than the phone's own, and the measurements came within plus or minus 1 dB of the reference. The authors attribute the improvement to removing the variability and limitations of built-in phone microphones. In other words, the hardware in your pocket is the limiting factor, not the software.

Question four: which weighting and which average?

A-weighting for anything you intend to compare with published figures.
C-weighting if the complaint is about low-frequency thud through a structure.
An equivalent continuous level over a stated period, not a peak.
The same choices for the before and the after readings.

An app that does not let you set these is an app that cannot be used for a comparison. Most free apps default to A-weighting and a fast response, which is fine for a first look and wrong for anything you intend to repeat, because a fast response reports the peaks and a comparison needs the average.

Question five: can you repeat it in a year?

Instruments drift, phones get replaced, and rooms change. If the measurement is meant to support a decision over time, ask what the calibration path is and what it costs. If there is no answer, treat the instrument as disposable and plan to re-baseline whenever it changes.

Question six: is measurement even the bottleneck?

Often it is not. If nobody disputes that the noise exists, measuring it changes nothing, and the money is better spent on the change itself. The ladder of options and what each rung buys is set out in what desk noise testing costs, and the stages of a full test are in what happens at each stage.

Common questions

Is the NIOSH app the best choice? It is the one with published accuracy figures and a stated test standard, which is why it appears in this cluster. It is iOS only.

Are Android apps unusable? Not unusable, but far less verifiable. Hardware differences between Android devices are exactly what the published testing could not control for.

How much does an external microphone cost? No figure is given here. Ask what it plugs into, whether it comes with a calibration certificate, and whether the app supports it.

Should I buy a calibrator too? Only if you own a meter and need results anyone else will rely on. For household comparisons it is unnecessary.