Calibration

A sound level meter converts a microphone voltage into a number. Calibration is what establishes that the conversion is correct, and without it the number is unanchored.

When acoustic measurement is part of a wider workplace review, how Microsoft Teams tracks activity can add operational context, but it should not be treated as a substitute for calibrated sound data.

What calibration actually is

Presenting the microphone with a sound of known level and adjusting the instrument to read it.

A field calibrator produces a reference tone — commonly 94 or 114 dB at 1 kHz — and fits over the microphone.

A useful external reference for further context is Brüel & Kjær sound knowledge.

The instrument is then set to display that value. Everything measured afterwards is relative to that anchor.

Without it you have a device reporting a number with no established relationship to the physical quantity, which is not a measurement in any useful sense.

Why phones fail here specifically

The microphone was designed for speech, with automatic gain control that adjusts sensitivity continuously — exactly the behaviour a measurement instrument must not have.

Sensitivity varies between handsets and between units of the same model.

And it changes with a case, a fingerprint, dust or a partially blocked port.

This is why only four of 192 apps met the accuracy criterion, and why the same four with an external calibrated microphone reached ±1 dB. The app was rarely the problem; the input was.

Field calibration and laboratory calibration

Field calibration is the check before and after a measurement session. Minutes, and it should be done every session — a reading whose calibration drifted mid-session is not recoverable afterwards.

Laboratory calibration is periodic verification of the whole instrument against traceable standards, typically annually, and it is what makes a meter's class claim meaningful.

One does not substitute for the other, and an instrument with a valid laboratory certificate still needs the field check.

What to do about drift

Check before and after.

If the post-session check differs from the pre-session one by more than a small margin, the session's data is suspect and the size of the discrepancy tells you how suspect.

Record both readings alongside the measurements. A measurement without its calibration record is not evidence for any purpose where evidence matters.

The proportionate version

For deciding whether a room is roughly quiet, loud or hazardous, an uncalibrated phone is adequate and knowing its limits is enough.

For anything that will be acted on — a workplace assessment, a complaint, a purchase decision — calibration is not optional, and the cost of a calibrator is small against the cost of an unusable measurement.

And for anything that could become a formal record, the instrument class and the calibration certificate are the substance rather than the number itself.

The short version

  • Calibration presents the microphone with a known level and anchors the instrument's readings to it
  • Without it, the device reports a number with no established relationship to the physical quantity
  • Phones fail because their microphones have automatic gain control, vary between units, and are affected by cases and dust
  • Only four of 192 apps met the accuracy criterion, and the same four with an external calibrated microphone reached ±1 dB — the input was the problem
  • Field calibration every session and laboratory calibration periodically; neither substitutes for the other
  • Record the before and after readings, because a measurement without its calibration record is not evidence