How Loud Is 120 Decibels? No Authority Defines This Number

We looked for a published definition of 120 decibels as a threshold — of pain, of immediate damage, of anything — in occupational regulation, standards documents and peer-reviewed hearing research. We did not find one.

What exists instead is a chain of citation between consumer articles, each quoting the previous one. That is exactly the pattern this library was rebuilt to remove, and the fact that the number is famous does not make it documented.

There is also a physiological reason for caution. Discomfort and pain depend on frequency content, duration, ear condition and individual sensitivity. A single broadband decibel figure cannot carry that, which may be precisely why no authority publishes one.

OSHA's noise standard handles very high levels in two ways. Its averaging table ends at 115 dBA for exposures of a quarter hour or less, and impulsive or impact noise is covered by a separate provision capping peak sound pressure level at 140 decibels.

In the vehicle world, UN Regulation 28 governs audible warning devices. A Class II device is tested on a bench, with the measurement made at 2 metres, and once fitted to a vehicle the assembly is checked against a minimum level measured 7 metres in front of it. Those are compliance figures produced by a specified test geometry.

Both examples make the same point differently: at high levels, the published record deals in peaks, ceilings and type-approval conditions — not in round landmark numbers.

At ordinary levels the difference between a peak and an average is a detail. At high levels it is the whole story. A gunshot or a firework can produce a peak far above any level that a shift average would ever reach, and it does so in a few milliseconds.

Capturing that requires an instrument with peak-capable detection and enough headroom to avoid clamping. Consumer microphone paths in phones do neither reliably, and browsers add automatic gain control on top.

That is why our meter is explicit about its own limits. Uncalibrated it reports an A-weighted digital level in dBFS, relative to digital full scale. After a one-time local calibration it reports an estimated dBA value. In neither state is it a peak-measuring instrument, and it should not be used to characterise impulsive events.

If your question is about a specific loud thing — a firearm, a firework, a siren, a horn — the useful path is the documented figure for that thing, with its distance and metric attached. Those figures exist, and they are far more informative than a landmark number.

If your question is about protection, the criteria that authorities actually publish are exposure criteria: a level paired with an averaging period, plus a separate peak ceiling for impulse. Those are the instruments regulators use, and they exist precisely because no single number does the job.