As an Amazon Associate, Akermin earns from qualifying purchases. Product links on this page are affiliate links.
A phone app is useful for comparing one situation to another and for spotting that something is very loud. It is not a measuring instrument in the sense that a sound level meter is. The difference is not the display, it is the microphone, the weighting filters, the time constants and the calibration chain. If you need a number you can put in a report, hand to a landlord, defend to an employer or compare against a legal limit, use a meter that conforms to IEC 61672 Class 2, such as the CRYSOUND CRY2831. If you are standing in a room wondering whether it is louder than yesterday, the phone in your pocket is fine.
What a phone microphone is actually built for
A phone microphone exists to make speech intelligible at arm’s length over a cellular codec. That design brief shapes everything about it. The capsule is tiny, the housing is sealed against water and dust, and the signal path is tuned by software to flatter the human voice and suppress wind and handling noise. None of that is a flaw for phone calls. All of it is a problem for measurement.
The first issue is frequency response. A measurement microphone is specified to hold a flat response across the audio band, so a 1 kHz tone and a 4 kHz tone of equal pressure produce the same reading. A phone microphone rolls off at the top and is often shaped in the low hundreds of hertz. The result is that two sounds of genuinely equal sound pressure level can read differently on an app, because the app is seeing a filtered version of the event. High-frequency noise, the hiss from a fan bearing or a compressed air leak, is exactly where a phone is weakest and exactly where the reading matters most.
The second issue is the A-weighting filter. Almost every noise limit you will encounter is expressed in dB(A), a curve that de-emphasises low frequencies to approximate how the ear responds at moderate levels. That curve is a defined mathematical filter, and it has to be applied to a signal that is itself uncoloured. Applying an A-weighting curve in software to a microphone that already has its own response curve gives you a number, but not a dB(A) value in the standards sense. The app cannot undo a distortion it does not know about.
The three things a meter has that an app cannot fake
Strip away the displays and the buttons and a sound level meter is three things a phone is not.
- A characterised microphone. The capsule is built and tested to a published response, and the meter’s electronics are matched to it. When the Extech 407730-NIST states 2 dB accuracy with 0.1 dB resolution, that figure is a claim about the whole instrument, capsule included, traceable to a calibration. The listing does not publish a frequency response tolerance, but the accuracy figure is the manufacturer’s statement about the instrument as a unit.
- Real weighting and time constants. A and C weighting are analogue or firmware filters applied to a known-flat signal. Fast and slow time constants are defined averaging windows, typically 125 ms and 1 s. An app can offer a switch labelled Fast or Slow, but it is averaging a signal it has already corrupted. The Extech 407730 publishes A and C weighting and a fast-responding analog bar graph, which lets you watch a transient peak that a numeric display would blur.
- A calibration path. A meter has a known reference point and an adjustment to bring it back to that point. You can attach an acoustic calibrator and check that the instrument still reads what it should. A phone has no such port and no such adjustment. This is the single biggest practical difference, and it is the one buyers underestimate.
The CRYSOUND CRY2831 is the clearest example of the category done properly: its listing states conformance to IEC 61672-1:2013 Class 2, a 29 dB to 138 dB range, 20 Hz to 12.5 kHz response and an IP54 rating. Those are the kind of published, checkable numbers an app cannot supply, because there is no standard an app is being measured against.
Why the accuracy claim on an app is not the number you think
App store listings often quote an accuracy figure, usually something like plus or minus a couple of decibels. Treat that with care. It typically describes the difference between the app’s reading and a reference meter under one specific condition, on one specific phone model, at one level and one frequency. Change the phone, the case, the grip, the temperature or the sound’s frequency content, and the figure no longer applies. The phone’s own microphone tolerance is usually the dominant error and is not published.
There is a second trap. Many apps let you type in an offset to match a reference meter. This is sometimes called calibrating the app. It is not calibration. It is a single-point correction that shifts the whole curve up or down. If the app reads 3 dB low at 1 kHz and 8 dB low at 4 kHz, a single offset cannot fix both. It makes the number at one frequency look right and leaves the rest wrong.
For context on what the weighting letters mean and how the curves differ, see what dB(A) means on a sound meter. The short version: the letter matters, and an app that does not apply the correct curve to an uncoloured signal is not giving you the letter it claims.
Where the phone genuinely wins
None of the above makes the app useless. It makes it a different tool. The phone wins on three counts.
Availability. It is already in your hand. For a first look at a noisy neighbour, a squealing pump or a room that feels too loud, the app gives you a rough magnitude and a trend. That is often enough to decide whether to escalate.
Logging and context. A phone can record a time-stamped series and a location, which a basic meter cannot. If your question is when did it get loud, not how loud exactly, the phone answers it.
Relative comparison. If the same phone, held the same way, reads 62 one evening and 74 the next, something changed. The absolute number may be wrong, but the difference is informative. This is the honest use of an app: as a comparator, not as a gauge.
Where the app loses outright is any situation with a consequence attached. A council noise complaint, a workplace exposure assessment, a tenancy dispute, a machinery warranty claim or a boundary measurement all require an instrument with a stated standard, a calibration history and a documented accuracy. A phone screenshot will not survive that scrutiny, and it should not.
Matching the instrument to the job
| Situation | Phone app adequate? | What you actually need |
|---|---|---|
| Is this room louder than it was yesterday? | Yes | Consistent phone, consistent hold, relative readings only |
| Rough idea of a machine’s noise level | Sometimes | A meter with stated accuracy, e.g. 2 dB as published for the Extech 407730-NIST and the Extech 130dB |
| Workplace exposure record | No | IEC 61672 Class 2 meter with calibration, e.g. The CRYSOUND CRY2831 |
| Noise complaint or tenancy dispute | No | Conforming meter plus a calibration certificate |
| Boundary or environmental assessment | No | Conforming meter, documented method, calibrated before and after |
| Checking a hearing protection decision | No | Meter with A and C weighting so you can compare the two |
Notice that the meter picks are not interchangeable either. The Extech 407730 and 407730-NIST publish 40 dB to 130 dB range and 2 dB accuracy. The CRYSOUND CRY2831 publishes a wider 29 dB to 138 dB range and explicit IEC 61672-1:2013 Class 2 conformance. If your work involves very quiet environments or very loud machinery, the wider range and the stated standard are the reasons to choose the latter. If your work is general spot checks in the middle of the range, the Extech units are the simpler instrument. One caution on the supplied data: the bullet text for the Extech Type Sound describes acceleration, velocity, displacement and RPM, which is vibration measurement, not sound. That text appears to belong to a different product, so the Extech Type Sound is not treated as a sound level meter here. If you want the broader buying picture, the comparison of sound level meters covers the field, and how to choose a sound level meter walks through range and standard selection.
The maintenance nobody mentions
A meter drifts. The microphone capsule ages, the diaphragm stiffens, dust settles on it and the electronics shift with temperature. This is normal and expected, which is why calibration is a routine, not an emergency. Most working meters are checked with an acoustic calibrator before and after a measurement session, and the two readings are recorded. If they differ, the session is suspect.
Phones have none of this. There is no calibrator that fits a phone microphone, no adjustment that survives a software update, and no record you can produce later. An app reading is a snapshot with no provenance.
Practical habits for meter owners: keep the capsule covered when not in use, avoid touching the diaphragm, let the instrument settle to ambient temperature before measuring, and note the calibration reading alongside every measurement. For a fuller treatment of the instrument itself, see what a sound level meter is.
One more honest limitation: a sound level meter measures sound pressure at the microphone. It does not tell you the source, the exposure duration, or whether the noise is harmful to a particular person. A meter is not an alarm and not a dosimeter. For personal exposure over a shift you need a dosimeter worn on the body, which is a different instrument again.
The short version
Use the app to notice. Use the meter to measure. The moment a number has to be defended, written down, compared to a limit or handed to somebody else, the phone has done its job and the meter takes over. Choose a conforming Class 2 instrument if the work is regulated, and a simpler stated-accuracy meter if it is not. Either way, calibrate it, and write the calibration down.
Frequently Asked Questions
Can a phone app measure decibels accurately enough for a noise complaint?
No. A complaint or dispute needs a reading traceable to a standard instrument with a calibration record. A phone app has no characterised microphone, no calibration path and no standard it conforms to. The reading may be close, but close is not defensible.
Why do two phones give different readings for the same sound?
Different microphone capsules, different housing acoustics, different software processing and different cases. Each phone's microphone has its own response curve, and none of them is published as a measurement specification. The difference between phones is usually larger than the difference between two conforming meters.
Is a cheaper meter with 2 dB accuracy good enough compared with a Class 2 meter?
It depends on the consequence. A 2 dB stated accuracy is fine for general spot checks and relative comparisons. For regulated work, environmental assessment or anything that will be challenged, the explicit IEC 61672-1:2013 Class 2 conformance and calibration certificate matter more than the headline accuracy figure. The listing for the CRYSOUND CRY2831 publishes that conformance; the Extech listings publish accuracy and range but not a stated IEC class.
Can I calibrate a phone app with an offset?
You can shift the reading, but that is not calibration. A single offset corrects one level at one frequency. It cannot correct a frequency-dependent error, and it does not create a calibration record. The app remains uncalibrated in any meaningful sense.
What does A weighting versus C weighting actually change?
A weighting rolls off low frequencies to approximate the ear at moderate levels and is the basis of most noise limits. C weighting keeps more low-frequency content and is useful for checking peak levels and low-frequency machinery. A meter that offers both, as the Extech 407730 listings state, lets you compare the two and see how much of the energy is low frequency.
Do I need a sound level meter at all if I just want to know if it is too loud?
Often not. If the question is simply whether a room feels loud, your ears and an app are enough. Buy a meter when you need a number with provenance: a record, a comparison against a limit, or evidence for somebody else. That is the line.