How Loud Should You Master? The Complete LUFS Guide

Every producer hits the same question right before release: how loud should my master be? Streaming changed the answer, but not in the way most online advice claims. The platforms turn everyone down to the same level, yet the pros are still mastering loud on purpose, and they're right to. Here's how loudness actually works now, and how to make the call for your own track.

I'm Tom Frampton. I run Mastering The Mix, and I master records professionally alongside building the tools. Over the years I've worked with artists including Tiësto, Calvin Harris, David Guetta, Camelphat, Steve Aoki and I have over 1,000 five-star reviews from real clients on my SoundBetter profile.

Loudness is the topic I see producers get most wrong, usually because they're following a streaming spec instead of trusting their ears and their mix. This page is the complete version of how loudness actually works, what the platforms do with it, and how to master loud the right way. Each section answers one question and links to a deeper guide if you want the full treatment.

1What is LUFS, and why does it matter?

LUFS stands for Loudness Units Full Scale. It's a measurement of perceived loudness, how loud a track actually feels to a human listener. That's different from a peak meter, which only shows the level of the single loudest sample. Two tracks can have identical peaks and feel completely different in loudness, and LUFS is the number that captures what your ears experience.

There are two readings you'll use. Integrated LUFS is the average loudness across the whole song, and it's the number streaming platforms measure to set playback level. Short-term LUFS is the loudness of the current few seconds, so it tells you how hard the loudest section, usually the drop or chorus, is actually hitting. When I talk about mastering to −6 or −9, I mean short-term LUFS at the loudest part of the track. When I talk about Spotify's −14 target, that's integrated. Keeping those two straight clears up most of the confusion around loudness.

To work with loudness you need to see it, not work blind. LEVELS shows your integrated LUFS, short-term LUFS, true peak, dynamic range and stereo information in real time, so you always know exactly where you sit. You can also set a target preset, Spotify, Loud, and others, and LEVELS warns you the moment you go over that loudness, so you get an instant visual cue when you've pushed past where you wanted to land. The wider explanation of mixing and mastering with these numbers is in mixing and mastering using LUFS.

2How loud should I actually master?

This is the real question, and the honest answer is that there isn't one number. There are lanes, and the track in front of you decides which one you're in. After years of mastering electronic music for DJs whose tracks need to hold up in clubs and on festival systems, here's how I actually frame the choice.

If you want the track to be loud and impactful, −6 LUFS short-term at the loudest section really is the sweet spot. It's hot, but not so hot that it introduces obvious distortion, and when the mix underneath is good it holds up and sounds great. This is what most of my clients ask for, and it's where the majority of commercial releases sit. Push much past −6 with a mix that's even slightly suboptimal and you start getting clipping in places, which just isn't professional.

If you'd rather optimise for dynamics and keep the transients fully open, go quieter, around −9 LUFS short-term, or even −10 to −11. At those levels the transients sound noticeably more open and less cut off. I wouldn't go much quieter than that though. There's no need, and if a track is too quiet, the platform may use a basic limiter you have no control over to bring it back up, which can do odd things to your sound.

Notice that the platform target and your mastering target are two different things. Spotify will normalise both a −6 and a −9 master down toward −14 integrated on playback, so the choice between loud and dynamic isn't about beating streaming, it's about how you want the track to feel and what the artist needs. Don't copy a number off someone else's meter. Pick the loudness that fits the music in front of you. The full breakdown of finding that balance is in how to get a loud punchy master and how to get the right amount of punch in your master.

3What does Spotify actually recommend?

Worth being precise here, because Spotify's own spec is more nuanced than the −14 figure most people repeat. There are two parts: the playback target, and the true peak ceiling, and the ceiling depends on how loud your master is.

Playback target: −14 LUFS integrated

This is Spotify's default normalisation level. Tracks mastered louder than −14 integrated get turned down to it on playback. Tracks quieter than −14 get turned up, but only with a limiter if the listener has the "Loud" volume setting enabled, which most don't. So your loud master and a quiet one both arrive near −14 for most listeners, regardless of where you mastered.

True peak ceiling: it depends on your loudness

This is the part people miss. Spotify gives two different true peak recommendations based on how hot your master is:

• Masters at or below −14 LUFS integrated: keep true peak below −1 dBTP.
• Masters louder than −14 LUFS integrated: keep true peak below −2 dBTP.

The reason is transcoding. When your lossless WAV gets encoded into Spotify's lossy formats (Ogg Vorbis and AAC), the process introduces inter-sample peaks. Loud, dense, brickwall-limited masters overshoot much harder during that conversion, so Spotify mandates the stricter −2 dBTP ceiling for loud music as a safety buffer. That extra dB of headroom stops the transcode overshoots from clipping the listener's converters and causing audible distortion. Anything over −14 LUFS integrated counts as "loud" by their definition.

The background on how Spotify's target settled at −14 in the first place is in Spotify's move to −14 LUFS.

4Why is loudness earned in the mix?

Here's the point that matters more than any target number: how loud you can go is decided in the mix, long before the limiter ever sees the track. Loudness on its own isn't the enemy. The cost only shows up in relation to the mix underneath it.

If the mix is good, where everything has its own space carved out, nothing's clashing, the instruments are well chosen and the balance is right, then you can push it loud and it stays clean. The better the mix, the louder you can take it before anything degrades. Flip that around and it gets ugly fast. Take a bass-heavy mix with a stack of overlapping frequencies, try to push it loud, and it just starts distorting. It sounds terrible, and there's no limiter setting that rescues it.

The squashing you hear in a crushed master is the same story. When a mix is full of overlapping frequencies, a loud master sounds crushed almost immediately. When things are carved out and separated, the same loudness sits much easier on the ears. So the single biggest thing you can do to master louder is to mix cleaner first. Two guides cover exactly that: controlling the low end where most of the damage happens, in how to control low-end distortion, and the wider clean-up in how to transform a messy mix into a pristine production.

5True peak, and why I turn it off

This one changed my mind after I actually ran the tests, so I'll be straight about it. True peak limiting clamps down on inter-sample peaks, the tiny level spikes that only appear when digital audio is converted back to analogue. The catch is that those peaks are inaudible on almost every playback system. So when you limit them, your limiter ends up working harder, sometimes two or three dB harder, and that extra work kills the transient punch you can actually hear, with no audible loudness gain to show for it.

Most mastering engineers I know work the same way. And here's the part people get backwards, including me until I tested it: true peak limiting only really makes sense on an optimised, dynamic master, not a loud one. A dynamic master has spare headroom, so there's room to apply the true peak limiting without it costing you much. A loud master has no room to give. Switch true peak limiting on over a −6 LUFS short-term master and you push it toward that −4 LUFS territory we covered in the Spotify section, the limiter clamps far harder, and the master sounds noticeably worse. So if anything, it's the loud master that can't absorb it, and the dynamic one that can. I'd still rather leave it off entirely and handle the true peak headroom per platform, which is what Spotify's −1 and −2 dBTP recommendations are really about.

So the two readings work together but you treat them differently: integrated LUFS is the loudness you're aiming for, and the true peak ceiling is a headroom decision you make per platform rather than a limiter you leave clamping the whole time. The exact-reading method is in how to master to get an exact true peak and LUFS reading.

6What did loudness normalisation really change?

For twenty years, mastering had a loudness arms race. Louder masters sounded more impressive in a head-to-head, so engineers pushed harder and harder. That was the Loudness War, and it ran from the mid-90s into the 2010s. Streaming was supposed to end it, because once every track gets normalised to the same playback level, being louder stops winning the comparison.

But here's what actually happened: almost nobody changed how they work. The top tracks on Spotify still average well above the −14 target, and labels still master loud. The war didn't end so much as go quiet. And the interesting thing is they're not wrong to keep mastering loud, because as we covered, a −6 master on a good mix sounds excellent, and most listeners would never hear it as crushed.

What normalisation genuinely changed is that the decision is now yours, made on purpose, instead of a race you're forced to run. And here's the reassuring part: when the mix is good and the processing is transparent, the difference between a loud master and a dynamic one, once they're both normalised to the same playback level, is tiny. Listen for yourself.

That's the whole argument in one A/B. With normalisation on, the loud and the dynamic master arrive at the listener at essentially the same loudness and sound almost identical, because the mix is good and the limiting is clean. Flip normalisation off and the only real difference is that the loud one is louder, the very thing the platform is about to undo. So there's no penalty for being a bit more dynamic, and no real prize for being the loudest. You're free to pick the loudness that suits the music. That's the real shift.

7How do I master loud without it sounding squashed?

Assuming the mix is already clean (and if it isn't, go back to section 4 first), squashing happens when you ask one limiter to do all the loudness work at once. You pull the ceiling down hard, the limiter grabs every transient, and the kick and snare lose their punch while the whole track gets that flat, lifeless wall-of-sound feeling. The fix is to spread the work out and to take out the problems before you push for level.

Clean the resonances first, so the limiter isn't wasting its reduction on one ringing frequency. Shape the tone and control the dynamics gently before the final limiting stage, so the limiter only has a little left to do. Then bring up the loudness in a way that preserves the transients instead of flattening them. Done in that order, you get level without the squash.

A limiter is still part of the chain, it's just not the whole chain. The right way to use one for mastering is covered in mastering the art of limiting and loudness, advanced limiting techniques and a guide to using a smart limiter.

8How do I hit an exact LUFS target?

This is where most of the slow, frustrating trial and error lives. The old way is to nudge a limiter, bounce, measure, find you're at −6.4 short-term, go back, nudge again, bounce again. You can spend twenty minutes creeping up on a number you should be able to just type in.

In FASTER MASTER's limit section you set the target directly. You pick the LUFS short-term max you want the loudest section to hit, choose whether you're optimising for streaming or loudness, and set the exact dB or dBTP ceiling yourself. You tell it how loud the master should be and it gets you there with the tone and punch intact, instead of you chasing the number by ear.

Here's the part that genuinely changes how you work: every adjustment you make to the EQ, transients, compression and stereo settings is gain-compensated in real time, so the loudness target stays locked no matter what you change. You can completely reshape the tone and the master stays at exactly the loudness you asked for, which means you're always comparing tonal changes at matched loudness instead of being fooled by level. That's incredible, and it removes the slowest, most error-prone part of mastering. It's also why the two normalised masters in the example above sound so close: the processing is transparent enough that loudness is the only thing really changing. The full walkthrough is in the ultimate guide to mastering with FASTER MASTER.

★★★★★

Armin van Buuren

Ranked #1 DJ by DJ Mag, a record of five times

"FASTER MASTER is perfect for when you need a quick, loud, and great-sounding master."

Try FASTER MASTER free →

9How do I check my loudness against a reference?

A number on a meter only tells you half the story. −6 LUFS short-term on your track and −6 on a commercial release can still feel different, because loudness perception depends on the tone and dynamics too. The way to know you're in the right place is to compare against a professional track in your genre, at matched level.

That last part matters more than anything. The louder of two tracks almost always sounds better in a quick comparison, even when it isn't actually better, so if you compare your master against a reference without matching their levels first, you'll fool yourself every time. Match the levels, then judge.

REFERENCE 3 lets you load commercial tracks, level-match them instantly, and compare your loudness, tone and width against them in real time. It turns "does this feel loud enough" into a direct, fair comparison. The method is in how to use reference tracks when mastering and the FASTER MASTER side in reference tracks in FASTER MASTER.

10How do I Quality Control the loudness before release?

Before the file leaves your computer, confirm the numbers you think you hit are the numbers you actually hit. Check the integrated LUFS across the whole track, not just the loudest drop. Check the true peak after you've accounted for codec encoding. Check that the loudness doesn't lurch between a quiet verse and a slammed chorus in a way that the platform will flag or that'll feel jarring on shuffle.

EXPOSE 2 is a standalone Quality Control application that scans the finished file and flags loudness and true-peak issues against release standards, so a problem you can't hear doesn't ship. It's the final gate before the master goes out. The full routine is in how to check your final masters like a mastering engineer.

11What's the workflow for a loud, clean master?

Here's the order that gets you loud without the squash, every time.

Do it in that order and loudness stops being a fight. You hit the target on purpose, you keep the dynamics that fit the track, and you confirm it before it ships. Loud and clean aren't opposites once you stop treating the limiter as the whole job, and once you accept that the mix did most of the work before the limiter even started.

?Loudness and LUFS FAQ