In this lesson
Loudness, PLR and LRA answer different questions
PLR, PSR and LRA all get called dynamic range. Learn what each one measures, how it is calculated, and which change moves which number.
· 8 min read · Lesson 32 of 44 in Mixing & Mastering
In short
- PLR and PSR measure how far the peaks stand above the loudness, while LRA measures how far the loudness moves between sections, so they answer for different kinds of dynamics.
- Read PSR section by section to find where the limiter flattens the hits, and read LRA over the whole song, never on a loop.
- Raise a low chorus PSR with the limiter and clipper settings, widen a narrow LRA with arrangement and automation, and compare both with a reference in the same style.
In this lesson (8)
Five readings, one verdict
A modern loudness meter at the end of the master chain shows a column of readings: integrated loudness, , , and sometimes . Most of us read the whole column as one verdict on "". Someone online says an LRA of 4 means the master is crushed, so you back the off, render again, and the LRA barely moves, because the limiter was never what set it.
The readings measure different things over different stretches of time. Read each one as the answer to its own question and they stop contradicting each other.
Why it matters
Two kinds of dynamics, two kinds of fix
Producers use "dynamic range" for two separate properties of a master: the micro dynamics of hits standing above the body inside a bar, and the macro dynamics of the song moving between sections, both described in the lesson on the final loudness push. Limiting and mostly change the first. Arrangement, automation and the balance between sections mostly change the second.
PLR and PSR read the first kind; EBU Tech 3343 calls PLR "a measure of micro-dynamics". LRA reads the second. EBU Tech 3342, the document that defines LRA, says it "should not be confused with other measures like dynamic range or ". Confuse them anyway and you fix the wrong thing: you soften a limiter to raise an LRA it hardly touched, or you see a healthy LRA and miss a chorus pressed flat against the ceiling.
The science
What each reading compares
All of these start from the ITU-R BS.1770 loudness measurement. They differ in what they compare and over how long.
PLR, the peak to loudness ratio, is the highest true peak in the whole file minus its integrated loudness, as the lesson on loud masters after normalization works through. It is one number for the whole song. Because most masters sit against a fixed true-peak ceiling, PLR mostly restates the integrated loudness: with a -1 ceiling, a -9 master has a PLR of 8 dB and a -12 LUFS master has 11 dB. One stray peak can also set it, however rare.
PSR, the peak to ratio, makes the same comparison against short-term loudness, the 3 second window, using the highest true peak inside that same window:
So PSR changes as the song plays, and it shows where the peaks have been pulled down toward the body, usually in the loudest chorus. Meters used to calculate it in slightly different ways, which is why an AES engineering brief proposed one shared definition (Shepherd et al., 2017).
LRA, the loudness range, measures how much the short-term loudness varies across the song. The meter takes 3 second readings at least ten times a second, drops every reading below -70 LUFS, then drops every reading more than 20 LU below the loudness of what is left. LRA is the spread of the remaining readings between the 10th and the 95th percentile, in LU:
The percentiles are there on purpose. A fade-out in the quietest 10% or a single huge hit in the loudest 5% cannot set the result alone. And because LRA compares 3 second averages, a limiter that shaves every snare by the same amount barely moves it. LRA falls when processing treats loud and quiet sections differently, which is what a hard final push does to a chorus.
A worked example ties the four together. Say the verses fill two thirds of a song at -13 LUFS short-term, the choruses fill the other third at -7 LUFS, and the chorus peaks reach -1 dBTP. Integrated loudness averages power, not decibels:
That is 1 dB louder than the plain average of the readings, because the loud sections weigh more. PLR is then 9 dB, the chorus PSR is 6 dB, and LRA is the 6 LU between verse and chorus.
The two kinds of dynamics have even moved apart across decades of releases. Deruty (2011) measured released music from 1969 to 2010 and found the crest factor falling sharply from around 1990, with no obvious decrease in LRA, as defined in Tech 3342, over the same years. A later peer-reviewed study of tracks from 1967 to 2011 reached the same split: the loudness war reduced how far the peaks stand out, but did not reduce long-term musical dynamics (Deruty and Tardieu, 2014).
The demo isolates the first kind. With normalization on, both loops play at the same loudness, so what you hear is how far the hits rise above the body, the distance PLR and PSR read. A loop has no verse and no chorus, so LRA would have nothing to report here.
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.
Loading demo…
Start with your speakers or headphones low.
Try it
Pull the readings apart
You need a loudness meter that shows LRA as well as integrated loudness, maximum true peak and short-term loudness. Some stock meters leave LRA out; free meters that follow EBU Tech 3342 show it.
0 of 7 steps done
Each change moved the reading that matches the kind of dynamics it touched. Step 6 is the one to watch: an LRA that shrinks under a harder limiter means the chorus is losing its lift over the verse.
Common mistake
Grading a master by one reading
A single reading makes a poor score. A high PLR can come from one snare hit in an otherwise flat master. A low LRA is normal for a club track built to hold one energy level, and it says nothing about whether the kick still punches. A wide LRA in a ballad does not prove its choruses are intact. Compare readings with references in the same style, and listen before you act on any of them.
The second mistake is reading LRA on a loop or a single section. With nothing to compare, the number means little, and Tech 3342 warns that very short programmes with silence at the start or end can read misleadingly high.
Takeaway
Pick the question first
Before you change anything, decide what you are asking. How loud will it play: integrated loudness. How much room do the peaks have across the whole file: PLR. Where is the limiter flattening the hits: PSR, section by section. How far does the song move: LRA.
I keep the two kinds of fix apart. Limiter and clipper settings are for the hits; arrangement and automation are for the sections. With normalization on, a streaming service matches the integrated loudness away, and a gain change keeps every distance the other three readings measure, so those distances are what the listener still hears.
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.
Loudness range
LRA, in LU: the spread between the quieter and louder parts of a programme, taken from the 10th to the 95th percentile of its gated short-term loudness (EBU Tech 3342). It tracks section-to-section change, not punch.
Peak to loudness ratio
The gap between a master's true peak and its integrated loudness. Heavy limiting makes it smaller. Plain gain normalization moves peak and loudness by the same amount, so PLR does not change: at a -14 LUFS target it is how far the loudest peak sits above -14. One stray peak can make it high on a flat master.
Peak to short-term loudness ratio
The gap between the highest true peak in a 3-second window and the short-term loudness of that window. It follows how far the peaks stand above the music moment by moment.
LU
Loudness Unit: a difference in loudness on the LUFS scale. 1 LU is the same step as 1 dB.
Limiter
A very fast compressor with a very high ratio. It stops peaks from going over a set ceiling.
Clipping
Cutting off the top of a waveform at a fixed level. Hard clipping flattens the peaks; soft clipping rounds them, which adds fewer harsh harmonics.
Crest factor
The gap between a signal's peak level and its average level. Heavy limiting makes it smaller.
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.
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.
Short-term loudness
Loudness measured over a sliding 3-second window, useful for comparing sections. Momentary loudness uses 400 ms.
Loudness normalization
A streaming service turning each track up or down so everything plays back at a similar loudness. A louder master is simply turned down further.
3 questions
Check yourself
Sources
Show all 6Hide sources
- Deruty, E. (2011). 'Dynamic range' & the loudness war. Sound On Sound, September 2011. www.soundonsound.com/sound-advice/dynamic-range-loudness-war
- Deruty, E., & Tardieu, D. (2014). About dynamic processing in mainstream music. Journal of the Audio Engineering Society, 62(1/2), 42-55.
- European Broadcasting Union. (2023). Tech 3342: Loudness range: A measure to supplement EBU R 128 loudness normalization (Version 4). EBU. tech.ebu.ch/docs/tech/tech3342.pdf
- European Broadcasting Union. (2023). Tech 3343: Guidelines for production of programmes in accordance with EBU R 128 (Version 4). EBU. tech.ebu.ch/docs/tech/tech3343.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/
- Shepherd, I., Grimm, E., Tapper, P., Kahsnitz, M., & Kerr, I. (2017). Measuring micro-dynamics, a first step: Standardizing PSR, the peak to short-term loudness ratio. Audio Engineering Society Convention 143, e-Brief 373.
Next lesson · 33 of 44
Streaming loudness myths waste mastersOne forum says master to -14 LUFS, the next says go much louder. Spotify documents how its normalization works, and the rules settle both arguments.Previous: Stop treating LUFS like a target
Written by Virzy Guns
Songwriter and producer, founder of Virzy Guns Production. Now building HealingWave.