Other meanings of Loudness normalization
Audio engineering
Loudness normalization is audio processing that adjusts perceived loudness to a standard level, allowing speech, music, advertisements, and programs to play at more consistent subjective volume. Unlike simple peak normalization, it estimates human-perceived loudness from the signal’s frequency content, duration, and channel arrangement.
Loudness normalization makes different audio programs sound more equally loud by measuring and adjusting perceived level rather than merely matching waveform peaks. The distinction matters because two recordings with identical peak amplitude can differ substantially in subjective loudness owing to spectral balance, dynamics, duration, and compression. Modern broadcast and production practice commonly expresses absolute loudness in LUFS, a unit associated with the measurement method standardized in ITU-R BS.1770.1
A normalizer usually calculates a program’s measured loudness, compares it with a target, and applies a gain offset. If a program measures below target, gain is added; if it measures above target, gain is reduced. The process does not necessarily change dynamics, although a subsequent limiter may be required to prevent clipping. The result is more predictable transitions between programs and less need for listeners to adjust volume manually.
Loudness measurement combines frequency weighting, channel summation, and time averaging to approximate important aspects of human hearing. ITU-R BS.1770 specifies a K-weighting filter and energy-based calculation, with separate momentary, short-term, and integrated readings; a relative gate prevents extended quiet passages from disproportionately lowering a program’s measured value.1
The EBU R 128 framework applies these principles to broadcast production and distribution, using an integrated target of −23 LUFS and a permitted tolerance for typical program delivery.2 Loudness units are relative: a change of 1 LU corresponds approximately to a 1 dB change in level. True-peak measurement is normally considered alongside loudness because digital samples can produce higher reconstructed peaks during digital-to-analog conversion.
Loudness normalization is used in broadcast television, radio, podcast production, streaming playback, game audio, and post-production to reduce abrupt level differences between adjacent content. In television, standards such as ATSC A/85 describe measurement, mixing, metadata, and operational practices for maintaining consistent loudness across the program chain. Streaming services may apply their own playback targets, so a master that sounds correct in one environment can be attenuated or left comparatively unchanged in another.
Normalization is not the same as dynamic-range compression. Gain normalization shifts an entire program, whereas compression reduces the difference between quiet and loud passages. A heavily compressed master may therefore remain fatiguing even after its average loudness is matched to a more dynamic recording. Loudness estimates also cannot perfectly predict every listener’s response, especially with headphones, small speakers, bass-heavy material, or speech mixed over music.
Loudness range and gating reveal why a single integrated value cannot describe an entire listening experience. Loudness range (LRA) summarizes meaningful variation within a program and can help engineers decide whether speech, music, or effects need adjustment without forcing every section to the same level.3 Gating also means that silence and very quiet atmospheres may contribute differently from sustained program material to the final reading.
Normalization can be performed before distribution, through delivery metadata, or during playback, and these approaches are not interchangeable. Metadata may be ignored, altered, or interpreted differently by receiving equipment; offline gain changes are more predictable but can expose true-peak problems. The standards are designed for program-level consistency, not for declaring one artistic loudness as universally correct. Deliberately quiet works, wide-dynamic classical recordings, and sound-design-heavy games may require contextual judgment even when all meet the same numerical target.
Targets and delivery practices vary by medium and platform; numerical loudness matching should be checked together with true peak, dynamics, and the requirements of the destination system.
Help improve the encyclopedia. Reports go straight to the site manager.