EBU R128 is the European Broadcasting Union recommendation that replaces peak-based level control with loudness normalization, setting a target of -23 LUFS integrated loudness, a maximum true peak of -1 dBTP and a descriptor called Loudness Range. In audio mastering and broadcast delivery, it defines how loud a finished program should measure, not how loud it should peak.
The recommendation was published in 2010 and revised several times since. It rests on the measurement algorithm of ITU-R BS.1770 and is accompanied by a set of technical documents that specify the meter, the Loudness Range calculation and practical production and distribution guidelines. Its practical effect is that programs, advertisements and trailers arrive at consistent perceived loudness across a channel.
What EBU R128 specifies
EBU R128 specifies three core parameters that every compliant program must report and meet: integrated loudness, maximum true peak level and Loudness Range. Integrated loudness describes the average perceived loudness of the whole program, measured from start to finish with gating, and is the value normalized to the target.
The following table summarizes the parameters, their target values and the role each one plays in delivery.
| Parameter | Target or limit | What it controls |
|---|---|---|
| Integrated loudness | -23.0 LUFS | Average loudness of the full program |
| Tolerance for file-based programs | Plus or minus 0.5 LU | Allowed deviation from the target |
| Tolerance for live programs | Plus or minus 1 LU | Deviation accepted when exact normalization is impractical |
| Maximum true peak | -1 dBTP | Headroom for inter-sample peaks and conversion |
| Loudness Range (LRA) | Reported, no fixed limit | Macro-dynamics of the program |
| Short-form content | Recommended short-term maximum near -18 LUFS | Perceived punch of commercials and promos |
The values are expressed in LUFS (loudness units relative to full scale) for absolute levels and in LU for differences, where one LU equals one decibel of level change. The Loudness Range has no mandatory limit because the appropriate range depends on genre, platform and listening environment.
How loudness is measured under EBU R128
Measurement under EBU R128 follows the K-weighted, gated algorithm of ITU-R BS.1770. Before any energy is summed, the signal passes a two-part weighting curve: a treble shelf that lifts upper frequencies to approximate how the head and ear canal favor them, and a low-frequency roll-off that keeps bass energy from dominating the reading. Each channel then contributes its mean square energy, with surround channels weighted slightly higher than front channels.
Gating keeps silence and soft passages from distorting the result. Blocks quieter than -70 LUFS are discarded first, and a second threshold set 10 LU under the provisional average then drops the remaining soft material, so the integrated figure describes what the audience actually hears in the foreground.
EBU Mode meters display several time scales, each suited to a different task, as the next table shows.
| Measurement | Integration window | Typical use |
|---|---|---|
| Momentary loudness | 400 ms sliding window | Fast visual feedback while mixing |
| Short-term loudness | 3 s sliding window | Judging sections and short-form maxima |
| Integrated loudness | Whole program, gated | Compliance with the -23 LUFS target |
| Loudness Range | Statistics of short-term values | Describing macro-dynamics |
| True peak | Oversampled sample values | Checking the -1 dBTP ceiling |
Loudness Range is calculated from the distribution of short-term values after a relative gate of -20 LU. It is the difference between the 10th and 95th percentiles, which excludes brief silences and isolated loud events such as a single explosion.
EBU R128 compared with other loudness standards
EBU R128 belongs to a family of BS.1770-based standards that share the same measurement method but differ in target values and regulatory status. The differences matter when one master is delivered to several regions or platforms.
| Specification | Region or context | Integrated target | Peak limit |
|---|---|---|---|
| EBU R128 | European broadcast | -23 LUFS | -1 dBTP |
| ATSC A/85 | North American television | -24 LKFS | -2 dBTP typical |
| ARIB TR-B32 | Japanese television | -24 LKFS | -1 dBTP |
| OP-59 | Australian television | -24 LKFS | -2 dBTP |
| Streaming music platforms | Online playback | Commonly -14 to -16 LUFS | -1 to -2 dBTP |
LKFS and LUFS denote the same unit under two names, so a 1 LU difference between European and North American targets is a genuine level offset, not a measurement artifact. Streaming services normalize playback rather than reject files, which is why music masters are usually far louder than broadcast targets and are simply turned down on playback.
Mastering and delivering to EBU R128
Mastering for EBU R128 means shaping the program around a fixed average level instead of pushing peaks toward full scale. Because the target is relatively low, a broadcast master typically retains several decibels more headroom than a commercial music master, and limiting can often be gentle or absent.
The decision table below lists common delivery situations and the approach that usually works.
| Situation | Recommended approach | Reason |
|---|---|---|
| Finished file-based program | Measure, then apply static gain to -23 LUFS | Static gain preserves the mix balance |
| Peaks exceed -1 dBTP after gain | Use a true-peak limiter with modest reduction | Prevents clipping in later conversion |
| Very wide Loudness Range | Apply gentle compression or manual rides | Quiet dialogue may become inaudible at home |
| Live broadcast | Mix to a short-term reading near target | Integrated value is unknown until the end |
| Commercial or promo | Watch short-term maximum as well | Dense short content can still sound aggressive |
Delivery for lossy codecs deserves extra care, because encoding can create new inter-sample peaks. Distribution guidance in the R128 family therefore suggests a lower peak ceiling, often around -2 dBTP or lower, for signals that will pass through data-reduced codecs.
Common misconceptions and pitfalls
Most problems with EBU R128 come from treating it as a peak or limiter setting rather than a measurement framework. The most frequent mistakes are listed below.
- Normalizing to sample peak: a file peaking at -1 dBFS can still measure anywhere from very quiet to very loud.
- Ignoring the gating: long pauses do not lower the integrated value as much as an ungated average would suggest.
- Over-compressing to hit the target: R128 rewards dynamics, so heavy limiting gives no loudness advantage after normalization.
- Confusing dBFS and dBTP: true peak includes inter-sample peaks, which ordinary sample meters miss.
- Assuming one master fits every platform: broadcast and streaming targets differ by 7 to 9 LU.
Stripped to the essentials
EBU R128 is an agreement among European broadcasters about how loud programs should sound. Before it existed, channels controlled only the highest peaks, so advertisements could be squeezed to sound much louder than the film around them.
The rule measures loudness the way ears hear it and asks every program to land on the same average value of -23 LUFS. Think of it as setting every guest at a dinner table to the same speaking volume. The result is simple: viewers stop reaching for the remote when a show changes to a commercial.
Where it is applied
A mastering engineer prepares a 50-minute documentary for a European public broadcaster. The mix measures -19.8 LUFS integrated, a true peak of -0.3 dBTP and a Loudness Range of 17 LU, because quiet interviews alternate with loud archival footage.
The engineer first rides the loudest archival scenes down by about 3 dB and lifts several soft interviews, which brings the Loudness Range to 12 LU. A static gain reduction of 3.2 dB then sets the program to -23.0 LUFS, and a true-peak limiter catches the remaining overs at -1 dBTP.
The delivered file passes the broadcaster’s automated quality control on the first attempt. On air, the documentary sits at the same perceived level as the news before it and the trailers after it, and the interviews remain clear on small television speakers.