LUFS · ITU-R BS.1770-4
LUFS is loudness measured the way you hear it.
Loudness Units relative to Full Scale. One number for a whole track, weighted for how human hearing responds to frequency, with the quiet parts excluded so a song is not judged by its intro.
It exists because peak meters were never a loudness measurement. Two tracks can peak at exactly the same level and be six decibels apart in how loud they sound, which is the whole reason the loudness war was winnable in the first place. LUFS closed that gap: it is the unit every published loudness target is written in, and the unit your master is delivered in.
What the standard specifies
What the standard defines.
K-weighting
A filter shaped like hearing
A high-shelf plus a high-pass, applied before anything is summed. Low bass contributes less to the reading than it does to a peak meter, because it contributes less to how loud a track feels. This is why a bass-heavy mix and a bright one can peak identically and read differently.
Gating
Silence does not count
Two gates. An absolute one at −70 LUFS discards true silence; a relative one discards everything more than 10 LU below the ungated average. Without them, a long quiet intro would drag the whole track's number down and a song would be measured by its least representative bar.
Integration window
Momentary, short-term, integrated
Momentary is 400 ms, short-term is 3 s, integrated is the whole programme. When a platform or a delivery spec says LUFS with no qualifier it means integrated — one number for the track. The other two are for watching while you work.
The scale
1 LU is 1 dB, and 0 is the ceiling
LUFS is an absolute scale referenced to full scale, so every value is negative and −14 is quieter than −9. A Loudness Unit is the same size as a decibel, which is why a difference between two readings is written in LU and a level is written in LUFS.
Not the same thing
LUFS, dBFS and RMS measure different questions.
Scrolls sideways →
| Unit | Answers | Frequency weighted | Used for |
|---|---|---|---|
| LUFS | How loud does this sound over the whole track? | Yes — K-weighted | Streaming normalisation, delivery specs |
| dBFS peak | What is the highest sample value? | No | Nothing much, on its own |
| dBTP (true peak) | What is the highest level between the samples? | No | Whether it will distort after encoding |
| RMS | What is the average energy? | No | A rough loudness proxy, superseded by LUFS |
The row worth understanding is the third. A digital file only stores sample values, but a converter reconstructs a continuous waveform between them, and that waveform can go higher than any sample in the file. A track can peak at exactly 0.0 dBFS and still clip a listener's converter, or a lossy encoder, at +0.8 dBTP. That is why every delivery target on this site pairs a loudness figure with a true-peak ceiling, and why the ceiling is the half that is never traded.
The loudness penalty
How much a platform turns your track down.
Normalisation is subtraction, and the arithmetic is not hidden. A platform that publishes a −14 LUFS target plays a −8 LUFS master back 6 LU quieter than you made it. That difference is what people mean by a loudness penalty.
Spotify and SoundCloud both state −14 LUFS in their own documentation, so the penalty on those two is your integrated reading minus −14, whenever your reading is louder. AES TD1008 states −16 for a track and −14 for an album. Apple, YouTube, TIDAL, Amazon Music and Deezer publish no loudness figure at all, so there is no penalty to calculate for them.
What normalisation does not undo
Being turned down is not the same as being made quiet, and a loud master is not “pointless” because normalisation exists. Turning a −8 LUFS master down by 6 LU does not produce a −14 LUFS master: the limiting that got it to −8 — the density, the reduced range, the way transients sit — is baked into the file and survives the gain change.
It is also not always on. Spotify's own documentation says normalisation is off on its web player and on third-party devices. You choose the loudness target.
The question underneath
What LUFS should you master to?
Every article that gives you one figure is answering a question you did not ask. −14 is what two platforms publish. It is not what a club record, a vinyl cut or a hip-hop release is mastered to, and it never was.
What is actually true: the loudness you choose is a decision about where your record sits next to the ones it will be heard beside, and the true-peak ceiling you choose is a technical fact about what happens to the file after you upload it. This is why the delivery matrix here is two axes rather than a list of platform names — naming a platform implies it published something, and most of them have not.
Read what each platform actually says How the chain hits a target
Questions
What people ask about LUFS.
What does LUFS stand for?
Loudness Units relative to Full Scale. LKFS — Loudness, K-weighted, relative to Full Scale — is the same measurement under a different name, used mostly in broadcast. A reading in LKFS and a reading in LUFS are directly comparable.
How many LUFS should my master be? Is −14 the standard?
It is what Spotify and SoundCloud publish, and what AES TD1008 recommends for an album. It is not a standard the whole industry agreed to, and most of the platforms it gets quoted for publish nothing at all. Treat it as one target among several rather than a rule.
What is the difference between LUFS and LU?
LUFS is a level and LU is a difference. Your track measures −9.4 LUFS; it is 4.6 LU louder than a −14 target. One LU is the same size as one decibel, so the arithmetic is straightforward.
Why do two meters give me different LUFS readings?
Usually because one is showing momentary or short-term rather than integrated, or because it is measuring a selection rather than the whole track. Compliant meters measuring the same integrated programme agree closely — the checker on this site was verified against ffmpeg's BS.1770 implementation and matched to within 0.15 LU.
Does a higher LUFS number mean better quality?
No. It means more of the dynamic range was spent on level. Whether that suits your record is a musical judgement, not a technical one — which is why this product gives you the target as a choice and then tells you exactly what it achieved.
Can I measure LUFS without buying anything?
Yes. The checker on this site reads integrated loudness, true peak and loudness range from a file in your browser, with no account and no upload — the file is decoded on the page and never sent anywhere.
One number, three meanings
The rules for measuring LUFS are published.
Read your own track's number.
Integrated loudness, true peak and range, measured in your browser. No account, no upload, no queue.