In this lesson
Stop treating LUFS like a target
A loudness meter tells you how loud a song measures, not how good it sounds. What integrated LUFS averages, and why chasing a reading flattens your drums.
Updated · 6 min read · Lesson 31 of 44 in Mixing & Mastering
In short
- Integrated LUFS is one energy average for the whole song, K-weighted, measured in 400 ms blocks, with very quiet passages gated out.
- Two masters can read the same LUFS and sound completely different, because the number says nothing about peaks, punch or how the loudness moves.
- Set the level by ear at matched loudness, then use the meter to check what you deliver.
In this lesson (8)
The number you were told to hit
You finish a mix, put a on the master and pull the threshold down while you watch the loudness meter. A tutorial said -8, so you stop when the display reads -8 . The number is right. The kick has lost its weight, the snare has stopped cracking, and the chorus no longer lifts out of the verse.
The meter did its job: it told you how loud the file measures. It never promised that a reading would make the master good. LUFS is a measurement, and treating it as a quality target pushes you to keep processing until the number arrives, whatever that costs the music.
Why it matters
A target changes what you listen for
With a target on screen, your attention moves from the drums to the display. You stop asking whether the chorus still hits and start asking how much more gain reduction you need. Every decibel of limiting past the point where the song sounds right is paid for in and in the contrast between sections.
The reward for that cost is smaller than it looks. Spotify measures the integrated loudness of each track and, on its default setting, turns anything louder than -14 LUFS down to -14 during playback. A master pushed to -8 LUFS plays at the same loudness as one left at -11. You keep the side effects of the limiting and lose the level you paid for.
Listen
Hear what normalization does to a loud master
A synth loop or a real mix, as a dynamic master and a loud, clipped one. Turn on streaming-style normalization and compare them at the same loudness.
Start with your speakers or headphones low.
The science
What the meter averages
Loudness meters follow ITU-R BS.1770. First the signal goes through K-weighting: a shelf that lifts the upper frequencies by about 4 dB to model the effect of the head, then a stage that reduces the weight of deep bass. The meter squares the weighted signal and averages it over blocks of 400 ms that overlap by 75%. For a stereo file the loudness of one block is:
Here and are the mean squares of the K-weighted left and right channels. The constant -0.691 cancels the gain of the K-weighting filter at 997 Hz. One , a loudness unit, is the same size as one decibel.
Integrated loudness then applies two gates. Blocks below -70 LUFS are ignored. The meter averages the rest, drops every block more than 10 LU below that average, and averages again. What is left is one number for the whole song.
That design explains what the number misses:
- It is an energy average. A snare crack lasting a few milliseconds is a tiny part of a 400 ms block. Limit those peaks off and raise the gain, and the reading goes up, because the body of every sound rose with the gain. The meter rewards the process that removes punch.
- It hides movement. One number cannot tell you how far the chorus rises over the verse. (a 3 second window) and momentary loudness (400 ms), defined in EBU Tech 3341, show how the level moves.
- It ignores peaks. Peak level is a separate measurement. BS.1770 defines it too, as .
Try it
Same number, different master
0 of 6 steps done
At the same integrated reading, the limited copy usually sounds flatter: the drums have less snap and the chorus rises less on the short-term meter. The numbers matched. The music did not.
Common mistake
One number for every song
Carrying one target across every genre and arrangement ignores what each song can take. A dense electronic track with sustained synths can take more limiting before it falls apart than a sparse acoustic song, where the space between notes is part of the sound. A number from a tutorial knows nothing about your song.
The opposite mistake is deciding that, because services normalize, -14 LUFS is now the rule. Spotify does suggest -14 LUFS integrated in its mastering tips, but nothing breaks when a master is louder. It is turned down, and it should keep its true peak below -2 ; the lesson on streaming loudness myths explains why.
Takeaway
Decide by ear, check with the meter
Decide how loud the master should be by listening at matched loudness. Push the limiter only while the song keeps its punch and its lift. Then read the meter to confirm what you are delivering: the integrated loudness, how far the short-term level moves, and a true peak under your ceiling. The meter is the right tool for checking a master and the wrong one for deciding how it should sound.
Limiter
A very fast compressor with a very high ratio. It stops peaks from going over a set ceiling.
LUFS
Loudness Units relative to Full Scale. A loudness measurement weighted toward how the ear hears. Streaming services use integrated LUFS, an average over the whole track, to match playback level.
Transient
The short, sharp start of a sound, like the crack of a snare or a pick hitting a string. It carries a lot of how hard a sound feels and how quickly the ear recognizes it.
High-pass filter
A filter that removes energy below a cutoff frequency and lets everything above it through.
LU
Loudness Unit: a difference in loudness on the LUFS scale. 1 LU is the same step as 1 dB.
Short-term loudness
Loudness measured over a sliding 3-second window, useful for comparing sections. Momentary loudness uses 400 ms.
True peak
The highest level a waveform reaches between samples once it is converted to analog or decoded from a lossy file. It can sit above the highest sample value, so meters estimate it by oversampling.
dBFS
Decibels relative to full scale, the largest value a fixed-point file or converter can hold, so levels below it are negative numbers. Inside a floating-point mix a signal can read above 0 dBFS and stay intact; it clips where it leaves float or passes through a plugin that clips. dBTP is the same scale measured as true peak, and it can read above 0.
3 questions
Check yourself
Sources
Show all 3Hide sources
- European Broadcasting Union. (2023). Tech 3341: Loudness metering: 'EBU Mode' metering to supplement EBU R 128 loudness normalisation. EBU. tech.ebu.ch/docs/tech/tech3341.pdf
- International Telecommunication Union. (2023). Recommendation ITU-R BS.1770-5: Algorithms to measure audio programme loudness and true-peak audio level. ITU. www.itu.int/rec/R-REC-BS.1770/
- Spotify for Artists. Loudness normalization on Spotify. support.spotify.com/artists/article/loudness-normalization/
Next lesson · 32 of 44
Loudness, PLR and LRA answer different questionsPLR, PSR and LRA all get called dynamic range. Learn what each one measures, how it is calculated, and which change moves which number.Previous: Saturation, clipping and limiting: three ways to get loud

Written by Virzy Guns
Songwriter and producer, founder of Virzy Guns Production. Now building HealingWave.