LUFS · ITU-R BS.1770
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.
Why a loudness unit exists
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 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 loudness of the blocks that passed the first gate. 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 louder is closer to 0
LUFS is an absolute scale referenced to full scale. Music in practice reads below 0 LUFS, so the numbers are 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.
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 | No | A rough loudness proxy, superseded by LUFS |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 its true peak can still reach +7.1 dBTP, which is what the checker measured on its own clipped-noise test file at that same 0.0 dBFS peak. 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.
How much a platform turns your track down
Turning a loud track down is subtraction, and the arithmetic is not hidden. Spotify, which normalises to −14 LUFS on its default setting, plays a −8 LUFS master back 6 LU quieter than you made it. That difference is what people mean by a loudness penalty.
Spotify states the −14 LUFS it normalises to, so its penalty is your integrated reading minus −14 whenever your reading is louder; it also turns a quieter track up, as far as the true peak allows with 1 dB to spare. SoundCloud recommends −14 and says it normalises, without stating the level. AES TD1008 recommends −16 for track normalisation and −14 for the loudest track of an album. Apple, YouTube, TIDAL, Amazon Music and Deezer publish no loudness figure for a stereo track, so there is no penalty to calculate for them.
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.
Read your own file's integrated loudness and true peak, in your browser, with no account and no upload.
Measure your own file, freeWhat LUFS should you master to?
Every article that gives you one figure is answering a question you did not ask. −14 is the figure Spotify and SoundCloud publish. It is a playback level, not a rule for how loud a club record, a vinyl cut or a hip-hop release should be.
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. Choose Dynamic (−14 LUFS), Balanced (−11), Full (−9), or Extra loud (−8). Each uses a −0.1 dBTP ceiling; lossy encodes can peak higher than the WAV.
Read what each platform actually says How the chain hits a target
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 the loudest track of 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 EBU’s compliance tests allow ±0.1 LU. The checker on this site is not certified: compared with ffmpeg’s BS.1770 meter on five 48 kHz test signals, it read within 0.25 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.