Skip to main content

Loudness metering, checked against the broadcast standard

BWS1 min readBYTHECLOUD ENT

BWS measures the loudness of every mix it exports. Its meter was run on test programmes whose answers come straight from the broadcast loudness standard.

The result

0.005LU

Largest difference from the standard’s answer

13

Test programmes

0.1LU

Tolerance the standard sets for meters

Source: Loudness meter against the standard · measured October 29, 2026

A mix delivered for broadcast or streaming is judged by one number: its integrated loudness. A meter that reads half a unit high or low makes every delivered mix wrong by the same amount. So the meter is checked against answers that do not come from the meter.

Across 13 test programmes the largest difference was 0.005 LU, and the mean difference 0.002 LU. The standard allows meters 0.1 LU.

How the answer is fixed

The standard publishes the coefficients of its frequency-weighting filters. From them the exact gain of a sine at any frequency follows in closed form, and the standard’s gating procedure, applied to the exact block energies of each programme, gives its loudness. The meter instead filters the samples and gates the blocks it measures itself.

The programmes cover four levels of a reference tone, six frequencies from low bass to the top of the audible range, and three sequences that exercise the relative and absolute gates with quiet passages around a louder one.

The signals are synthesised, so the run is the same every time. The check covers the meter’s weighting and gating; it does not cover true-peak measurement or loudness range.

More from AI research notes from Vancouver

  1. StorMStorM 0.111.0
  2. StorM Web EngineWhere the sun is, checked against an ephemeris
  3. StorM Web EngineThe StorM Web Engine initial load