Reading loudspeaker specifications: the conditions behind the numbers
Abstract. An SPL figure without its measurement conditions is not a specification. This note explains the conventions professional data sheets use for sensitivity, maximum SPL, power ratings and subwoofer figures, and how to compare numbers between manufacturers without being misled.
Sensitivity and the one-watt convention
Sensitivity states the sound pressure level at one metre for one watt of input, averaged across the enclosure's pass band. It is the anchor for every other level figure: maximum SPL is, to first order, sensitivity plus ten times the base-ten logarithm of peak power. A specification whose maximum SPL exceeds that ceiling is claiming output its own numbers cannot support, and the arithmetic takes thirty seconds to check.
Crest factor, programme and peak
Power ratings differ by the signal used. A continuous rating reflects pink noise held for hours; programme is conventionally twice continuous, and peak four times, corresponding to a crest factor of four, which is representative of dynamic music material. Maximum SPL figures on SSOUNDS sheets are peak levels at one metre with pink noise at crest factor four, driven to the clipping point of the recommended amplifier with the system preset applied. When comparing sheets, confirm which convention each manufacturer used before comparing the numbers.
Free field and half space
Full-range enclosures are quoted under free-field conditions: no reflecting surfaces. Subwoofers are conventionally quoted in half space, on a reflecting plane, because that is how they are deployed; the boundary adds approximately 6 dB relative to free field. The two conventions are both legitimate and both in industry use, but a half-space subwoofer figure compared against a free-field figure flatters the former by the boundary gain. SSOUNDS sheets mark half-space figures explicitly.
What a data sheet cannot tell you
Directivity data on any manufacturer's sheet describes a single cabinet under the stated conditions, and system behaviour in a venue depends on the array, the geometry and the room. Prediction files for acoustic modelling software carry measurement provenance for exactly this reason. SSOUNDS issues prediction data files from measured polar data only; where charts on our documents are computed from the driver complement, the method is stated on the document itself.