−27 LKFS and −23 LUFS are four units apart and are not interchangeable, because the gating and the channel weighting behind the two figures are specified by different documents.
Two loudness targets that are not the same instruction
written 2026-07-29rots yes — figures above moveresponsible Room Tone Register (a named register, not a person)
The two documents
| Document | Target | Gate | Channel weighting |
|---|---|---|---|
| EBU R 128 | −23 LUFS | −70 LUFS absolute, −10 LU relative | K-weighting per BS.1770 |
| Streaming mix specification | −27 LKFS ±2 LU, dialogue-gated | Dialogue-gated, not programme-gated | K-weighting per BS.1770 |
Both build on the same loudness algorithm, and both use the same K-weighting. The difference is what gets measured. A programme-gated measurement includes everything above the gate; a dialogue-gated measurement anchors on the dialogue and lets music and effects sit where they like around it. A mix tuned to hit one will not hit the other by construction, because the two measurements are of different things.
Why a preset fails
A loudness normalisation preset carries a target and a gate. Copy a −23 LUFS programme-gated preset into a workflow that expects a dialogue-gated −27 LKFS delivery and the preset will do exactly what it says and produce the wrong result: it will pull the whole programme to a target that was never the specification, and it will not tell you, because the number it reports is the number you asked for.
What the register does instead
The register does not mix. It records the two specifications side by side and states the difference, because the failure mode is a person with two correct-looking numbers and no way to tell which one the delivery actually wants. When in doubt, read the specification document for the specific delivery, not a summary of it, and note that these figures are reissued — which is why this page carries a checked date and is marked as a figure that rots.
A true-peak limit is a ceiling, not a target. Setting the mix to sit at the ceiling is a different decision from setting it to hit the loudness target, and the two decisions are frequently conflated in delivery notes.
Source log
- 01Complexity of Exact Reverberation, CCRMA (Stanford)supports the computational cost of an exact reverberant model, used here as an analogy for method-dependent targets
- 02About the Reverberation document, CCRMA (Stanford)supports the framing of the document set behind the loudness figures
- 03Feedback Delay Networks for Late Reverberation, CCRMA (Stanford)supports the late-field treatment that a dialogue-gated measurement resembles
Each line points at one specific document. No line is a home page.
Leave this page Close this page if you need a recording level rather than a delivery target — the delivery target is a post-production decision.