What LUFS measures
LUFS (Loudness Units relative to Full Scale) is perceived loudness, standardized. Unlike a peak meter, which reports the highest sample, a loudness meter filters the signal through K-weighting - an EQ curve approximating the ear's sensitivity, which discounts deep bass and emphasizes upper mids - then averages energy over time and gates out silence. The result tracks what your ears report: a sustained dense chorus reads loud; a single sharp transient barely moves it.
Three windows matter: momentary (400 ms, the "now" reading), short-term (3 s, good for section-by-section comparison), and integrated (the whole program - the number platforms normalize on). One loudness unit (LU) corresponds to one decibel of gain, so "3 LU too loud" means the platform will apply −3 dB.
Its partner metric is true peak (dBTP): the estimated analog peak between samples after reconstruction. Platforms ask for −1 dBTP ceilings because lossy encoding pushes inter-sample peaks over the top of a 0 dBFS master, producing audible crackle on playback.
Platform targets
| Platform | Target | True peak | Notes |
|---|---|---|---|
| Spotify | -14 LUFS | -1 dBTP | Premium users can pick -19/-14/-11 in settings; quiet tracks are only boosted if limiting allows |
| YouTube | -14 LUFS | -1 dBTP | Turns loud content down, never boosts quiet content |
| Apple Music | -16 LUFS | -1 dBTP | Sound Check on by default on iOS; the gentlest common target |
| Amazon Music | -14 LUFS | -2 dBTP | Slightly stricter true-peak recommendation |
| Tidal | -14 LUFS | -1 dBTP | Album normalization: preserves track-to-track dynamics within a release |
| Deezer | -15 LUFS | -1 dBTP | Splits the difference between Spotify and Apple |
| SoundCloud | none | -1 dBTP advised | No loudness normalization - the last big holdout where loud masters still play louder |
| Broadcast (EBU R128) | -23 LUFS | -1 dBTP | The standard the streaming targets descend from; podcasts commonly use -16 to -19 |
Targets as commonly published mid-2026; platforms adjust occasionally - treat ±1 LU as noise.
The sane mastering strategy
Don't master to −14. The targets are playback levels, not delivery specs. A master that sounds right at −9 LUFS gets turned down to −14 and keeps all its character; a master strangled to exactly −14 to "avoid being turned down" gains nothing and often sacrifices punch. The only numbers to actually respect are the true-peak ceiling (−1 dBTP) and your own genre's expectations for density.
Compare loudness-matched. The classic self-deception is A/B-ing your master against a reference without matching levels - louder always sounds "better" for the first seconds. Match integrated LUFS between your master and the reference before judging tone and punch; every serious metering plugin does this in one click.
Check the quiet-track edge case. Most platforms turn loud tracks down but do not boost quiet ones fully (or at all). Acoustic and classical material mastered very quiet (−20 LUFS and below) can end up noticeably quieter than everything else in a playlist - the one case where nudging toward the target genuinely helps.
Your monitoring chain matters more than the meter: level-match your speakers and set a consistent reference volume first - the studio monitor setup guide covers it, the dB converter handles the unit math, and the decibel meter gets your room level in the right zone.
LUFS & Streaming Loudness FAQ
What is LUFS?
Loudness Units relative to Full Scale - a standardized measure of perceived loudness, not just signal level. It applies K-weighting (a filter approximating how ears weight frequencies) and gates out silence, so it tracks how loud material feels over time. One LU equals one dB of gain change. "Integrated LUFS" measures a whole track; short-term (3 s) and momentary (400 ms) windows track sections.
What LUFS should I master to for Spotify?
Spotify normalizes to -14 LUFS integrated with -1 dBTP true peak headroom - but that does not mean you should master to exactly -14. Master the music to sound right; if that lands at -10 LUFS, Spotify simply turns it down 4 dB and nothing is lost except meaningless meter numbers. What you should avoid is crushing dynamics purely to hit a loud number, because after normalization the crushed master plays at the same volume as a dynamic one - only flatter.
What is the difference between LUFS and dB?
dBFS measures instantaneous signal level against digital full scale - a peak meter. LUFS measures perceived loudness over time with frequency weighting and gating. Two tracks can both peak at -1 dBFS while one is 10 LU louder than the other. Loudness normalization works on LUFS precisely because peak level says almost nothing about how loud something feels.
What is true peak (dBTP) and why -1?
Digital samples are points on a curve; the reconstructed analog waveform between samples can swing higher than any sample - an inter-sample peak. True peak metering estimates that reconstructed maximum. Platforms recommend ceilings of -1 dBTP (or -2 for lossy-heavy delivery) because lossy encoding (AAC, Ogg) pushes peaks higher, and a 0 dBFS master can clip audibly after encoding.
Is the loudness war over?
Mostly, on platforms that normalize - a -8 LUFS brickwall master gains nothing there and loses punch. It survives in pockets: SoundCloud does not normalize, DJs mixing tracks back-to-back compare raw levels, and normalization can be disabled by users. The practical takeaway: dynamics are free again on the big platforms, so let the mix breathe and stop chasing numbers.
How do I measure LUFS?
Any modern DAW ships a loudness meter (Logic, Ableton 12, Reaper's JS meters, Cubase all include one), and free plugins like Youlean Loudness Meter measure integrated/short-term/momentary LUFS plus true peak. Measure the whole track for integrated LUFS - the number platforms act on - and check true peak stays under the ceiling.