Decibel Comparison: Which Figures Are Actually Documented

Before any comparison is possible, a figure has to declare what it is. In the sources reviewed for this library, five families cover almost everything.

Regulatory declaration: a manufacturer files a value under labelling law, usually sound power in dB(A) re 1 pW, measured to a test code. Type approval: a product must meet a level under a specified test geometry, such as a horn measured at 2 metres on a bench. Exposure criterion: an authority pairs a level with an averaging period and an exchange rate. Dosimetry: an instrument worn by a person records an accumulated exposure over a shift. Peer-reviewed measurement: researchers publish levels with their microphone positions and conditions stated.

A number from one family cannot be moved into another. That is the whole reason charts that mix them go wrong, no matter how carefully the individual figures were copied.

Earlier versions of this page carried a colour-coded ladder running from a whisper to a jet engine, with a single number for each. None of those numbers had a traceable publisher, a stated distance or a stated quantity.

They have been removed rather than adjusted. Adjusting them would have meant inventing better-looking values, which is the same failure with more effort. The chart is shorter now and every row can be checked.

The colour coding went with them. A colour gradient across a scale implies that positions on the scale are commensurable and that the boundaries mean something. Neither is true when the underlying figures come from different families.

Four questions settle almost every ambiguity. Which quantity is this — sound power or sound pressure? At what distance, or under what mounting and test geometry? With which frequency weighting, A, C or Z? And over what period — an instant, a peak, a shift average, a 24-hour average?

A figure that answers all four can be compared with another figure that answers all four in the same way. A figure that answers none of them is decoration.

This is not pedantry. It is the difference between a dishwasher label and a kitchen measurement, between a horn bench test and a street, between an 8-hour criterion and a moment on a display.

Your own measurements belong to none of these families, and that is fine as long as it is stated. Uncalibrated, our meter reports an A-weighted digital signal level in dBFS — relative to digital full scale, not to sound pressure. After a one-time local calibration against a trusted reference reading, it reports an estimated dBA value on consumer hardware.

That makes it genuinely useful for comparisons: this room against that room, window open against closed, machine on against off. It does not make it a source you could cite in a chart like the one above, and we do not pretend otherwise.