How to Read a Loudness Meter: LUFS in Your Mix

A loudness meter gives you several numbers at once, and each one answers a different question. Short-term tells you how your sections sit against each other, integrated tells you how the whole song measures, and true peak tells you whether the file will clip after encoding. Read the right number at the right moment and the meter becomes a very useful second opinion. Read the wrong one and it will quietly send you in circles.

I've watched plenty of producers stare at a flickering LUFS display, unsure whether the number they're looking at is good news or bad. So let's go through it properly: what each reading means, which to watch while mixing, how to set the meter up, and what to do with the numbers once you have them.

TL;DR

• A loudness meter shows five readings: momentary (last 400 ms), short-term (last 3 seconds), integrated (whole track), true peak and loudness range.
• Use short-term while you mix to compare sections, like verse against chorus. Use integrated only as a final delivery check.
• Put the meter last in the chain, reset it before you play the song from the top, and let it run through at least the first chorus before trusting it.
• 1 LU is the same size as 1 dB, so moving your master by 2 dB moves the reading by 2 LU.
• Treat the meter as a supplement to your ears, not a boss.

Table of Contents

• What a loudness meter is actually measuring
• The five readings, one by one
• Setting up your meter so the numbers are honest
• What numbers should you be looking at?
• Worked example: reading one song start to finish
• Common mistakes when reading a loudness meter
• Our perspective: loudness is the start of the conversation
• Getting the next step from your numbers
• Quick reference: reading your meter in one pass
• FAQ
• Sources

What a loudness meter is actually measuring

LUFS stands for Loudness Units relative to Full Scale. It is a measure of perceived loudness, not raw signal level. LUFS and LKFS are effectively the same thing, both K-weighted loudness relative to digital full scale, and one loudness unit equals a 1 dB change in level.

The K-weighting is the clever part. Our ears are more sensitive in the mid-range than at the extremes, and LUFS accounts for that, unlike older RMS or VU style metering, which can be blind to frequency-dependent loudness differences. That is why a bass-heavy mix and a bright mix can show the same peak level on a sample meter yet feel completely different in loudness, and why LUFS usually matches what you hear more closely.

It does not replace listening, though. The same AES guidance stresses that loudness metering should supplement your mixing decisions and not dictate them. Keep that in mind and the rest of this article will make a lot more sense.

The five readings, one by one

Most modern meters show the same five things. Here is how I think of each.

The windows come from the BeamTracer explanation of the five readings, which lists momentary at 400 ms and short-term at 3 seconds.

Momentary: the twitchy one

Momentary moves fast. It is great for noticing that a snare hit or a vocal shout jumps out, but it will make you chase ghosts if you try to mix to it. If you find yourself nudging faders because the M number moved, step back. That reading is designed to wobble.

Short-term: your mixing companion

Short-term is the one I would keep on screen the whole time you are mixing. Because it averages over three seconds, it follows the musical phrase rather than every transient. Industry experts recommend using short-term to check how loud song sections are relative to each other, such as the chorus against the verse, and saving integrated for the final delivery check. Watching the wrong meter for the task is one of the most common ways producers get misled about their dynamics.

A practical example: you play your verse and short-term hovers around the same value as your chorus. The chorus will probably feel flat, however much you've EQ'd it. Maybe the verse is too hot, maybe the chorus needs more energy from layers, width or a lift in the arrangement. The meter has pointed at the problem and your ears decide the fix.

Integrated: the whole-song verdict

Integrated averages everything from the moment you reset it. It is the number streaming platforms use for loudness normalisation, so it is the one to quote when someone asks how loud your master is.

Two behaviours catch people out:

1. It is cumulative. If you don't reset before a new pass, the meter keeps averaging the old audio in. The best practice is to reset just before the first playback of the entire song you want to measure.
2. It uses gating. Very quiet material (below -70 LUFS) and sections 10 LU quieter than the running average are ignored, according to iZotope's guide. So the reading will keep shifting early on. Let it run through at least the first chorus before you put real weight on it.

If you want a deeper walk through, we've written about LUFS from a producer's point of view on our blog.

True peak: the clipping safety net

True peak, shown in dBTP, estimates the peaks that can occur between samples when your audio is converted to analogue or encoded to a lossy format. A file can show 0 dBFS at the sample level and still produce distortion downstream. We cover this in more detail in our true peak guide for mastering and in the piece on intersample peaks and -1 dBTP headroom.

Loudness range: how much your song breathes

LRA is a statistical measure of how much the loudness varies across the track. A low number suggests a steady, compressed track, a higher one suggests more contrast between quiet and loud passages. It isn't good or bad on its own, it depends on the genre and your intention. Our post on loudness range and why it matters for mastering goes further.

Setting up your meter so the numbers are honest

A perfectly good meter can give you rubbish readings if it sits in the wrong place. Here is the setup I'd follow every time.

1. Put it last. For an accurate project-level reading, the meter belongs post-fader at the very end of the plugin chain. If your DAW can't do post-fader inserts, route everything to a dedicated master bus where the meter is the final plugin, and keep that fader at 0 dB throughout the measurement.
2. Switch off anything that changes level after it. A limiter or gain stage after the meter means the meter is lying to you.
3. Reset before you play. Go to the start of the song, hit reset, press play.
4. Play the whole song, uninterrupted. Skipping around or looping a section gives you the loudness of what you played, not of the song.
5. Read short-term during, integrated after. Note the integrated number at the end, together with the highest true peak.

Some meters have an automatic gain adjustment feature that helps you reach a target level. That is handy for mastering, but while mixing I would rather keep gain changes under my own hands so I understand what I'm doing.

If you want to compare what several tools do, ones built into your DAW, plugin-based, or an audio workspace such as a browser-based sound studio for mixing and fine-tuning audio, the same rule applies: know which window each number is using before you compare them.

What numbers should you be looking at?

This is where most searches for "how loud should a mix be" end up, and the honest answer is: it depends on what stage you're at.

Streaming targets

Spotify normalises playback toward -14 LUFS, and its artist guidance says masters should keep true peak below -1 dBTP, or -2 dBTP for masters louder than -14 LUFS, to minimise extra distortion during transcoding. The same page notes that the web player and third-party devices such as TVs or speakers don't apply normalisation, so playback can differ by device.

Broadcast uses a different, considerably lower target than streaming, so never carry a streaming master straight into a TV or film delivery without checking the spec you've been given. For our take on current numbers, see our streaming loudness targets post and how we meter for streaming.

Don't chase the target blindly

Mastering engineer Ian Shepherd points out that platforms normalise to -14 LUFS, but the song's artistic quality matters more than hitting that figure exactly. He suggests avoiding levels consistently above -9 LUFS to protect musicality, while acknowledging that many pop hits are mastered considerably louder, as covered on mastering.com. A louder master doesn't earn you a louder playback on a normalising platform, it just gets turned down, often with less punch than a more dynamic version.

And before mastering?

This is the question behind "mix level before mastering" and "pre master loudness". My view is that your mix doesn't need to hit any final loudness at all. What it needs is:

• Healthy headroom, so that the mastering stage isn't fighting a mix that already peaks at the ceiling.
• Balanced short-term readings from section to section.
• No limiter on the master bus that you're using as a crutch.

So don't ask your mix to be -14 LUFS integrated. Ask whether it is balanced, clean and open enough that mastering can bring it to its target without damaging it. If your mix feels timid next to your references, we wrote about how to measure first and then fix a mix that's too quiet.

Worked example: reading one song start to finish

Let's say you've bounced a mid-tempo pop track. Here is how a calm reading session goes (the numbers are made up for illustration).

1. You place the meter last on the master bus and reset it.
2. During the first verse, short-term sits around -17 LUFS. Good, there is room to grow.
3. The first chorus pushes short-term up to roughly -13. That's a 4 LU lift, which will feel like a decent step.
4. The bridge dips to -18. The contrast feels natural.
5. The last chorus only reaches -13.5, which is lower than the first one. That is the meter flagging that the final chorus doesn't feel bigger than the first, so you go back to the arrangement and automation.
6. After the final note you read integrated, say -15.2, and true peak at -1.4 dBTP.

If your target were Spotify's -14, you'd know you have about 1.2 LU to find in mastering, and the peak is already safe. If you pushed the whole thing up in a limiter just to hit the number, you'd lose the 4 LU contrast that made the chorus work.

The point is that the first thing the short-term reading told you was about the music, not about compliance.

Common mistakes when reading a loudness meter

• Using integrated while mixing. It moves too slowly to guide fader decisions, and it needs the full song to mean anything.
• Forgetting to reset. Old audio drags the average, and you'll swear your mix changed when it didn't.
• Measuring a looped chorus. A chorus alone will always read louder than the song as a whole.
• Judging on the first 20 seconds. Gating means the early integrated value is unstable.
• Comparing numbers from different meters without checking the settings. Time windows, gating and true peak oversampling all need to match.
• Treating the target as pass or fail. Normalisation turns tracks up or down, it doesn't reject them.
• Ignoring the room and monitors. The meter can't hear a boomy room, and your ears can.

Our perspective: loudness is the start of the conversation

We built AubioMix because a loudness number on its own is just a number. Telling a producer their integrated reading is -12.8 LUFS says nothing about whether the compression is pumping, whether the low end is masking the kick, or whether the stereo image collapses in mono.

So when we look at loudness, we always want it in context. Our founder is a multi-award winning mix engineer with 20+ years of experience, and the way our reports work reflects that: they pair objective measurement with subjective guidance, so you see the figure and also what it means for your genre and your mix. Our evaluation framework is updated regularly to reflect current genre standards, which matters because a good loudness figure for deep house isn't the same as one for jazz or worship music. You can see what we mean on our genre benchmark pages, for example pop.

Our honest advice: use meters constantly, trust them for what they measure, and never let them overrule a mix that sounds right. A meter tells you where you are. You decide where you're going.

— AubioMix

Getting the next step from your numbers

Once you can read the meter, the useful question is what to change. If you'd like a second pair of eyes on a mix, you can upload it to AubioMix and get a report that covers loudness alongside compression, EQ, frequency masking, stereo width, phase, clipping and device translation, across more than 15 areas, each with specific actions to take. A single mix review costs £4.99, and the Professional plan is £19.99 per month if you review mixes regularly. If you only want to check a version against a reference, a reference pass is £2.99. Pricing details are on the pricing page, and the glossary is a handy place to look up terms like LUFS, true peak, crest factor and LRA while you read your report.

Quick reference: reading your meter in one pass

• Mixing a section? Watch short-term.
• Spotting a sudden spike? Glance at momentary, but don't chase it.
• Checking the final master? Reset, play the whole song, read integrated.
• Exporting? Check true peak against your platform's guidance.
• Wondering about dynamics? Look at LRA, in the context of your genre.
• Still unsure? Trust your ears, then use the meter to confirm.

If you'd like help with the vocabulary, our mixing terms explained simply page is a good companion.

FAQ

What is the difference between short-term and integrated LUFS?

Short-term averages the last 3 seconds and moves with the music, so it suits comparing sections while you mix. Integrated averages the whole song from the last reset and is the figure platforms use for normalisation. Use short-term to make creative decisions and integrated to confirm the finished master.

Why does my integrated LUFS keep changing at the start of a song?

Integrated loudness is cumulative and uses gating, so it ignores very quiet passages and sections far below the running average. Early in the song there is little audio to average, so the number drifts. Let the meter run through at least the first chorus before trusting it, and always reset before playback.

How loud should a mix be before mastering?

There is no fixed number, because your mix doesn't need to match the final target. Aim for sensible headroom, balanced short-term levels across sections and no limiter doing the heavy lifting on the master bus. Your mastering stage can then bring it up to the delivery level cleanly.

How many dB is 1 LUFS?

A change of 1 LU equals a change of 1 dB, according to Waves. If your master reads -15 LUFS and you raise it by 1 dB, it reads -14 LUFS. LUFS itself is an absolute scale relative to digital full scale, so it isn't a dB figure you can convert directly to sound pressure.

Sources

• How to Read a Loudness Meter: LUFS in Plain Language, BeamTracer
• Loudness normalization on Spotify
• Loudness Metering Explained, Waves
• Spotify LUFS, mastering.com
• Loudness basics, AES
• LUFS Meter Manual, Klangfreund
• What are LUFS? The complete guide, iZotope

Recommended

• LUFS explained: the producer's guide to loudness
• −14 LUFS, −1 dBTP: How Producers Meter for Streaming with AubioMix
• What is loudness range LRA, and why does it matter for mastering?
• Measure First, Then Fix a Mix That's Too Quiet for Producers