One ITU formula underlies every streaming loudness target

The system

ITU-R Recommendation BS.1770, published by the International Telecommunication Union's Radiocommunication Sector, specifies an algorithm to measure two things: the subjective loudness of an audio programme and its true-peak signal level. The current text, BS.1770-5, states its own scope plainly: 'This Recommendation specifies audio measurement algorithms for the purpose of determining subjective programme loudness, and true-peak signal level.' The ITU is a standards body, not a broadcaster or streaming platform, and the recommendation defines a measurement method rather than any single service's target loudness level.

What the documents establish

The ITU's recommendation page shows the document has been revised six times since its first version in July 2006, with the current BS.1770-5 dated November 2023. The recommendation's own text defines loudness as a K-weighted measurement: audio passes through a two-stage filter, one stage compensating for head-related acoustic effects and a second applying what the document calls RLB weighting, before a formula converts the result into a unit the document names LKFS, for 'Loudness, K-weighted, relative to nominal full scale.' The specification adds a two-stage gating process, first discarding quiet passages below an absolute threshold of -70 LKFS, then applying a second, relative threshold, so that silence does not skew a programme's measured loudness. A separate annex defines true-peak measurement in decibels relative to full scale, denoted dB TP, using oversampling of at least 192 kHz to catch inter-sample peaks that ordinary sample-value metering misses. Notably, the ITU document itself uses only the LKFS unit name; it does not use the term LUFS, which is the European Broadcasting Union's own name for the numerically identical unit in its related EBU R128 recommendation.

Craft and rights

For a mastering engineer, BS.1770 is the arithmetic underneath a decision that platforms make separately: what target loudness to normalize incoming audio toward. The recommendation supplies a shared, repeatable way to measure loudness; it does not itself set a target such as the specific LUFS values streaming services publish in their own delivery specifications. Confusing the measurement standard with any one platform's target is a common error worth avoiding: a track can be correctly measured under BS.1770 and still be turned up or down by a platform's own normalization to reach that platform's separately chosen target. There is no rights question of consent or licensing here; the relevant question for a producer is technical accuracy and knowing which target applies where a mix is delivered.

Outcomes and open questions

The recommendation continues to be revised, most recently to address object-based and channel-plus-object audio formats in newer annexes, meaning a workflow built around an older revision's channel assumptions may need rechecking against the current text as delivery formats change.

• Which BS.1770 revision, and which annex, does a given loudness meter or plug-in actually implement?
• Is a quoted loudness target a property of the ITU measurement standard, or a separate platform policy built on top of it?
• Does a workflow's gating and true-peak handling match the two-stage method the recommendation specifies?

A measurement standard is not a loudness policy; BS.1770 supplies the ruler, and platforms decide separately what mark on it they want.