How loud is too loud?

Two quantities, and you need both

• LAeq — the A-weighted continuous sound level over a time window, usually 15 or 30 minutes. This is the quantity that wears on hearing across an evening. A single loud moment barely registers in it; three hours consistently too loud certainly does.
• LCpeak — the C-weighted peak level. It describes a SINGLE event: the snare hit, the feedback squeal, the microphone knocked over. That kind of thing can damage hearing in milliseconds, no matter how quiet the evening was on average.
• DIN 15905-5 names both, and for good reason: limit only the continuous level and the bang gets through. Limit only the peak and you are not controlling the continuous level at all — you are just making the system quieter across the board and taking its dynamics away.

Why a number without a measurement means nothing

A digital level is not sound pressure. Between the number in the software and what arrives at the seat sit the amplifier gain, the loudspeaker sensitivity, the distance and the room. The same file produces 88 dB in a parish hall and 96 dB with the same system in a church — or the other way round.

So every honest sound pressure figure needs exactly one reference from outside: a sound calibrator on the microphone capsule, or a comparison against a sound level meter. There is no way around it. Software that promises a limit in dB(A) without anyone having held a measuring device is guessing — and a guessed limit is more dangerous than none, because people rely on it.

The mistake almost everyone makes — AURIX included

AURIX got this wrong itself until 7 August 2026. The hearing protection limited the unweighted peak while the control was labelled in dB(A). Measured: control set to 100 dB, output 82.3 dB LAeq. And the calibration was not measured but assumed — the software took it for granted that the measurement sweep had produced 90 dB.

This is written down here because the same mistake sits in many products of this class, and because you can spot it: ask which quantity is being limited and where the calibration comes from. If nowhere says how the device learned the sound pressure of your system, then it never learned it.

What you can actually do

Measure once. A sound calibrator costs around 150 euros and takes five minutes to fit; a phone app and a sound level meter are less accurate but still a measurement rather than an assumption. Stand where people will actually be listening — not at the desk, not in front of the speaker.

Then keep measuring while it runs. A reading from the start of the evening says little; the continuous level builds over hours. If a 15-minute window is only two minutes old, it is a two-minute figure — even if it says "LAeq,15min" above it.