What LUFS should you master to for Spotify, Apple Music and YouTube

Master to about -14 LUFS integrated, with true peaks no higher than -1 dBTP, and send the same file everywhere. That is the level at which Spotify and YouTube stop turning a track down, and nobody rejects a master for landing a decibel either side of it. Going louder than -14 buys you no extra volume on any of the three services with normalisation switched on. Going much quieter can cost you volume on YouTube and sometimes on Spotify. Apple publishes no loudness figure at all; the -16 LUFS number attached to Apple Music comes from outside measurement, not from Apple.

What LUFS measures, and which reading to use

LUFS stands for loudness units relative to full scale. It comes from the ITU-R BS.1770 standard and weights the signal to match how loud people hear it, so a bass-heavy track and a bright one at the same LUFS sound roughly as loud as each other. A peak meter cannot tell you that. A track peaking at 0 dBFS can be quiet, and a heavily limited one peaking at -1 dBFS can be very loud.

Most meters show three LUFS readings. EBU Tech 3341 defines them: momentary uses a 400 millisecond window, short-term a 3 second window, and integrated covers the whole stretch from when you press start to when you stop, with gating that ignores near-silence. Streaming services normalise on integrated loudness, so that is the number to read, and it only means something once the meter has heard the entire track. Watching the short-term figure bounce around in the chorus tells you very little about what Spotify will do.

True peak, written dBTP, is the other number. It estimates the peaks that appear between samples once the file is converted back to audio or encoded to a lossy format. A master whose samples never pass -0.1 dBFS can still overshoot 0 dB after encoding, which is why the streaming guidance is written in true peak rather than sample peak.

What each service actually publishes

The three companies document this very differently. Spotify publishes exact figures. Apple publishes a description with no numbers. YouTube publishes nothing, and the figures everyone quotes come from reading its player.

Spotify

Spotify's loudness normalization page is the clearest document of the three. It measures to ITU 1770 and adjusts tracks to -14 LUFS by default. Premium listeners can pick Loud at -11, Normal at -14 or Quiet at -19. Louder masters get negative gain. Quieter ones get positive gain, but only as far as the peaks allow, since Spotify leaves 1 dB of headroom for lossy encoding. Its own example is a track at -20 LUFS with a true peak of -5 dBFS, which gets lifted to -16 LUFS rather than -14.

Three details from that page are worth knowing. On the Loud setting Spotify runs quiet, dynamic tracks through its own limiter, set to engage at -1 dB with a 5 millisecond attack and 100 millisecond decay, so a gentle master can end up limited by Spotify instead of by you. When someone listens to an album start to finish, Spotify normalises the whole album at once, keeping the level differences between tracks; shuffle adjusts each track separately. And the web player and third-party devices such as TVs and smart speakers apply no normalisation at all. Spotify's guidance to artists is to target -14 LUFS integrated, keep true peak below -1 dBTP, and send one master per track.

Apple Music

Apple's loudness control is Sound Check. The Apple Digital Masters document says it measures each track's loudness, stores that in metadata, and uses it to raise or lower playback volume, and that it can work per album so deliberate level changes between tracks survive. It says tracks mastered loud will play back quieter and can sound weaker for it, and tells engineers to master for the sound they want regardless of playback volume. It gives no LUFS target. What it does ask for is 24-bit source files and no clipping, and Apple distributes a free command-line tool, afclip, to find clipped samples.

The -16 figure comes from mastering engineer Ian Shepherd, who reported in March 2022 that Apple had moved Sound Check to LUFS measurement, was normalising to -16 LUFS in line with the AES TD1008 streaming recommendations, and had switched Sound Check on by default. Treat it as a well-sourced measurement that Apple could change without saying so. For a stereo master it changes nothing about what you should do: a track at -14 LUFS simply comes out about 2 dB quieter on Apple Music than on Spotify, as does every other track around it. Apple does publish a hard figure for Dolby Atmos deliveries, -18 LKFS, which our Dolby Atmos guide covers.

YouTube

YouTube has never published a loudness target. Right-click any video, open Stats for Nerds, and the Volume / Normalized line shows how much the player has turned it down, plus a content loudness figure. Ian Shepherd's breakdown of that panel explains that a reading of 6 dB means the video is 6 dB louder than YouTube's reference, and gets a 50% volume reduction to compensate. Engineers reading that panel put YouTube's reference at about -14 LUFS. The part that matters most for mastering is the other half of the finding: YouTube does not turn quieter videos up. A master at -18 LUFS stays at -18 LUFS, 4 dB below everything that hit the reference.

Three masters, worked through

The easiest way to see what normalisation does is to run three versions of the same track through it: a loud master at -8 LUFS, one at -14, and a gentle one at -18 LUFS whose true peaks already reach -1 dBTP. The Apple column uses the reported -16 figure and should be read as approximate.

On the phone and desktop apps, the -8 and -14 masters end up at the same level, and the -8 one has had its transients and dynamics limited away to get there. That is the whole argument against loudness-war mastering on streaming: the platform removes the volume you gained and leaves the side effects. Spotify's own page adds that masters louder than -14 need a lower true peak ceiling of -2 dBTP, because hot masters distort more when encoded.

The bottom row is the honest counter-argument. On Spotify's web player and on third-party devices the -8 master really is louder, and the same goes for DJ software, Bandcamp downloads and a USB stick in a club. If most of your listeners hear you in those places, a denser master has a real reason to exist. The -18 master is the one with a clear problem: on YouTube and Spotify's Normal setting it sits 4 dB below its neighbours, and on YouTube there is no setting a listener can change to fix that.

How to measure your master

You need a meter that reads integrated LUFS and true peak, and what your DAW gives you out of the box varies a lot.

• Logic Pro has one built in. Apple's Loudness Meter guide lists momentary, short-term and integrated readings, a loudness range figure, and a target line you can drag anywhere between -30 and 0 LUFS.
• Ableton Live does not. We found no mention of a loudness or LUFS meter anywhere in Ableton's Live 12 release notes, so Live users need a plugin.
• FL Studio's Wave Candy has an ITU-R BS.1770 meter mode, but Image-Line's manual describes it as a long-window average and lists no separate integrated or true peak readout, so a dedicated meter is still worth adding.
• Youlean Loudness Meter 2 is free in VST, VST3, AU and AAX formats on Mac and Windows, and its free version reads integrated loudness, true peak, loudness range and dynamic range. The Pro version adds streaming-service presets and was $29 on offer against a $49 list price when we checked on 14 September 2026. Our free VST plugins guide covers it alongside other free mixing tools.
• Loudness Penalty, run by MeterPlugs, is a free website that takes a finished file and shows how much YouTube, Spotify, Tidal, Apple, Amazon, Pandora and Deezer are likely to turn it down.

A mastering routine that covers all three

• Finish the mix first. A limiter pushed into a muddy low end makes it louder and no clearer, and our mix engineer explainer covers when to hand that stage to someone else.
• Put the loudness meter last on the master bus, after the limiter, and set the limiter's ceiling to -1 dBTP with true peak or oversampling mode on.
• Play the whole song from the start with the meter reset, and read the integrated LUFS at the end.
• If it reads below about -14, bring the limiter input up until it reaches roughly -14, or stop earlier if the limiter starts audibly squashing the drums.
• If it reads above -14, decide whether you like how it sounds. If you do, drop the ceiling to -2 dBTP and leave it. If you pushed it there only to be loud, back off.
• Export a 24-bit WAV at the session's sample rate and upload that one file to your distributor for every service.

AI mastering services make these same decisions for you. Our look at LANDR, eMastered and BandLab compares what they cost. Whichever route you take, run the finished file through a meter before it goes anywhere.