LUFS, PLR and LRA explained in plain words
A loudness meter shows five or six numbers, and each one answers its own question: how loud the track is on average, how loud it is right now, how alive it is and whether it’s hitting the ceiling. Let’s go through them one by one — without formulas, but precisely enough to read an analyzer report without getting confused.
In short
- LUFS is the unit of perceived loudness per the ITU-R BS.1770 standard; 1 LU equals 1 dB.
- Integrated is the average loudness of the whole track, the one platforms use; Short-term covers 3 seconds, Momentary 0.4 seconds.
- LRA shows the spread of loudness between the quiet and loud parts of a track.
- PLR is the difference between the peak and the average loudness: how dense the master is. PSR is the same, but in the moment.
- Normal values for streaming: around −14 LUFS, PLR 8–12 LU, LRA most often 4–10 LU.
Contents
What LUFS is and why it beats peaks and RMS
LUFS stands for Loudness Units relative to Full Scale. It’s a way to measure loudness the way a person hears it, not the way a level meter sees it. The standard is described in the ITU-R BS.1770 recommendation, and the European broadcast standard EBU R128 added measurement rules to it. In American documents the same quantity is called LKFS.
How LUFS differs from the familiar meters:
- Peak level shows only the highest point of the wave. A hi-hat click and a loud chorus can have the same peak.
- RMS averages the signal’s energy, but treats all frequencies the same. A bass with lots of energy inflates the reading, even though the ear doesn’t perceive it as loud.
- LUFS first passes the signal through a K-filter: it attenuates the very low end and slightly boosts the top, as hearing does. Only then does it average.
One more useful property: 1 LU is 1 dB. If a track is at −9 LUFS and the target is −14 LUFS, the adjustment is exactly −5 dB.
Integrated, Short-term and Momentary
These are the same loudness measured over windows of different length. The shorter the window, the faster the number reacts.
| Measure | Window | What it’s for |
|---|---|---|
| Integrated | the whole track | The main number. Platforms normalize loudness by it. Quiet pauses don’t count — gating takes care of that. |
| Short-term | 3 seconds | The loudness of a phrase or a bar. It shows how much louder the chorus is than the verse. Short-term max is the loudest stretch of the track. |
| Momentary | 400 ms | The instant loudness of a single hit or syllable. Momentary max is the loudest short moment. |
Gating is the rule that stops silence from dragging the result down. Stretches quieter than −70 LUFS are discarded immediately, and then everything more than 10 LU below the preliminary average is discarded too. That’s why a long quiet intro barely affects Integrated — and therefore how the platform will normalize the track.
LRA: loudness range
LRA (Loudness Range) shows how much the quiet and loud parts of a track differ. It’s calculated from Short-term: the loudness distribution over the whole track is taken, the quietest 10% and the loudest 5% are discarded, and LRA is the difference between the remaining edges. That way one sudden hit or a quiet ending doesn’t skew the result.
- 2–4 LU — almost even loudness: dense pop, EDM, many hip-hop tracks.
- 4–10 LU — the most common range: a noticeable difference between verse and chorus.
- 10 LU and more — strong contrasts: ballads, rock with quiet bridges, orchestral music.
A small LRA isn’t a mistake in itself — it’s a matter of style. But if LRA is close to zero in a genre that needs a dramatic arc, the loud and quiet parts have most likely been squashed into one plane. The Stream-Q report compares LRA with tracks in the same genre.
PLR and PSR: how dense the master is
PLR (Peak to Loudness Ratio) is the difference between the maximum true peak and the integrated loudness. If the peak is −1 dBTP and Integrated is −9 LUFS, PLR is 8 LU. The smaller the number, the harder the track is limited: the peaks are cut and the average level is pulled up to the ceiling.
PSR (Peak to Short-term Ratio) is the same thing in the moment: the peak over the last three seconds minus Short-term. PSR shows exactly where the track is “flattened” — for example, only in the chorus.
| Value | PLR for the whole track | PSR in the moment |
|---|---|---|
| Very dense | 6 LU or less | under 6 LU |
| Dense but alive | 7–9 LU | 6–8 LU |
| Lively dynamics | 8–12 LU | 8–12 LU |
| Very dynamic | over 12 LU | over 12 LU |
Why this matters for streaming: after normalization all tracks sound roughly equally loud, and the one with more difference between the hit and the background wins. A track with a PLR of 11 LU at the same loudness feels bigger and more powerful than one with a PLR of 5 LU.
True peak: the last measure
True peak is the peak level taking into account what happens between samples. The meter oversamples the signal (at least four times, per the standard) and finds peaks a regular meter doesn’t see. For streaming, the target is no higher than −1 dBTP. More on inter-sample peaks and how to remove them in the article “Clipping and true peak”.
How to read these numbers together
A few typical combinations and what they usually mean:
- −7 LUFS, PLR 5 LU, LRA 3 LU — a very loud, dense master. Platforms will turn it down by 7 dB, and it may sound smaller than its neighbors.
- −11 LUFS, PLR 9 LU, LRA 6 LU — a typical modern master with its dynamics kept.
- −19 LUFS, PLR 14 LU — a quiet, dynamic track. Good for acoustic music, but on YouTube and in stream players it may stay quieter than others.
- −10 LUFS, true peak +0.8 dBTP — the loudness is fine, but the peaks are above zero: there will be distortion on encoding.
On Stream-Q streams the same numbers are shown by the loudness meter overlay — viewers see the Short-term scale, the histogram and the loudness history in sync with the player. The artist gets them in the report after the stream, together with conclusions in plain words.