Measuring noise with a phone microphone

A phone microphone produces a signal level, not a sound pressure level. Turning one into the other needs a constant that depends on the specific microphone, its position on the body, the case around it and whatever processing the operating system applies on the way. That constant is why two apps on two phones can disagree by ten decibels in the same room.

Where the number comes from

The app measures the signal's root mean square over a short window, takes twenty times its logarithm, and adds an offset chosen so that ordinary rooms land in a believable range. Everything except that offset is exact. The offset is an assumption, so treat a phone reading as a well-informed estimate rather than a measurement.

Two ceilings you cannot argue with

Phone microphones are built for speech and clip well before a real hazard. Anything past roughly 100 dB flattens out, so a phone systematically under-reads exactly where accuracy would matter most — at a concert, next to a saw, near machinery. Automatic gain control is the second problem: the system quietly turns the input down when things get loud, which flattens the curve further.

What phone measurements are genuinely good for

Comparison. Same phone, same spot, same position: which of these two fans is louder, does the neighbour's noise really get worse after eleven, is the office quieter with the door shut. Relative differences survive an unknown offset, and this is exactly the question most people are actually asking.