How to Read a Loudness Meter: LUFS in Plain Language
A loudness meter shows five readings: momentary (M) over the last 400 ms, short-term (S) over the last 3 seconds, integrated (I) across the whole track, true peak in dBTP, and loudness range (LRA). Integrated is the number streaming platforms judge. Read all five free on the Loudness tab of BeamTracer's Audio Analysis Engine, in Chrome or Edge.
Hi, I'm Trina with BeamTracer.com — let's put names on the five numbers your loudness meter keeps showing you.
At some point someone told you your track needs to "hit −14 LUFS", you opened a loudness meter to check, and it answered with five different numbers, none of which was labelled "your track". M, S, I, dBTP, LRA. A graph. A histogram.
Here's the thing: every one of those readings answers a genuinely different question, and once you know which question each one answers, a loudness meter stops being a cockpit and starts being the most honest instrument in your session. That's the whole post. No manual required.
What are the five numbers on a loudness meter?
| Reading | Window | The question it answers |
|---|---|---|
| M (momentary) | last 400 ms | How loud is it right now? |
| S (short-term) | last 3 s | How loud is this section? |
| I (integrated) | everything since reset | How loud is the whole track — the number platforms judge |
| True peak (dBTP) | inter-sample | Will it clip after encoding? |
| LRA | whole session | How much does the loudness move? |
Everything below is those five rows, expanded.
What does LUFS measure, and how is it different from RMS?
LUFS is a time-averaged loudness measurement, like RMS, with two fixes that make the average match your ears. If you grew up watching an RMS meter, you already understand LUFS — both average the signal's level over time instead of chasing peaks.
LUFS is frequency-weighted. Your ears don't hear all frequencies as equally loud, so the measurement runs through a filter curve (called K-weighting) that tips the scale toward what humans actually perceive — a bit more sensitivity up top, less rumble counted down low.
LUFS is gated. Silence and near-silence get thrown out of the average. An RMS reading of a track with a long quiet intro is dragged down by the intro; the LUFS integrated reading ignores what's effectively not programme. That's why LUFS is the number every streaming platform settled on.
LUFS, LKFS, LU — same thing
LUFS and LKFS are the identical unit with two names — the broadcast standard says LKFS, the music world says LUFS. And 1 LU = 1 dB, exactly. If a reading moves by 2 LU, the level moved by 2 dB. There is no conversion to memorize.
What do momentary, short-term and integrated loudness mean?
They are three clocks on the same signal: momentary averages the last 400 milliseconds, short-term the last 3 seconds, and integrated everything since the last reset. Integrated is the one platforms act on; momentary and short-term are what you watch while the music plays.
Momentary is the last 400 milliseconds — it jumps with every drum hit, and it's the number you watch when something spikes.
Short-term is the last 3 seconds — the loudness of the section you're in. When people say "the chorus is 4 dB hotter than the verse", short-term is the number they compared.
Integrated is the gated average of everything since you last hit reset. This is the one Spotify, YouTube and Apple use to decide how much to turn your track up or down — and it's meaningless for the first few seconds, so give it at least a chorus before you believe it.
All three clocks at once, live: the bright line is momentary, the filled area is short-term, the dashed line is the target — and the histogram underneath is quietly filing every second away.
One habit that will save you from bad readings: integrated loudness is cumulative, so reset before you measure. If you start a measurement mid-track, or leave the meter running across three different songs, the integrated number describes that whole accumulated mess, not your track.
What is true peak, and why is it not the peak in your file?
True peak (dBTP) is the highest level the reconstructed analog waveform reaches between your samples — which can be higher than any sample in the file. A sample meter reads the samples; true peak oversamples the signal the way the standard specifies and reads what the listener's converter will actually produce.
A normal peak meter reads the highest sample in your file. But the analog waveform your listener hears is reconstructed between those samples, and it can swing higher than any of them — a file that reads −0.1 dBFS on a sample meter can genuinely exceed 0 in reconstruction.
This matters because of what happens after you upload: lossy encoders (AAC, Ogg, MP3) reshape the waveform slightly and can push peaks up to a dB or so higher. A master that "never clips" on a sample meter can distort on the listener's phone.
The one number that is actually a rule
Loudness targets are negotiable — true peak isn't. Keep your master at or below −1 dBTP (−2 dBTP for very loud masters) and the encoded versions survive every platform. This is the closest thing to a law in this whole subject.
What does LRA (loudness range) tell you?
Loudness range measures how spread out your short-term loudness is over the session, in LU. A dynamic live set might read 15 LU; a brickwalled master reads 3–5. Neither is "correct" — LRA is a description, not a grade — but it tells you instantly which kind of track you're looking at.
One caveat the meter should tell you itself: LRA is statistical, so it needs about a minute of programme before the number means anything. A good meter flags the reading as settling until then instead of showing you a confident wrong answer.
The distribution histogram under the graph is the fastest way to see LRA: every second of short-term loudness files into a bin, so over a session it becomes the honest answer to "where does this mix live?" A twin-peaked histogram is a set with two levels. A single narrow spike is a master with one. The dashed brackets show the range the LRA calculation actually keeps — the extremes outside them are deliberately discarded by the standard.
If dynamics are the part you care about, there's a whole companion post on PSR, PLR and LRA — including the one threshold worth memorizing.
Can you measure a whole file without playing it through?
Yes. Drop an audio file onto the Loudness tab, or click ANALYZE FILE, and BeamTracer measures the entire track on your machine in a couple of seconds — integrated, true peak, LRA and the full history, always from start to finish. It is the reliable way to get an integrated number, because an offline pass can never be contaminated by a mid-track start or a forgotten reset.
Is there a free loudness meter that runs without installing anything?
Yes — BeamTracer's Loudness tab is a browser tab in Chrome or Edge, with every view, platform preset, file analysis and export free. Loudness metering used to mean a plugin purchase, and the free tiers ration exactly the parts you want. For comparison, as of this writing:
| Youlean Loudness Meter | DAW stock meters | BeamTracer Loudness | |
|---|---|---|---|
| M / S / I, true peak, LRA | Yes (free) | Yes, in most modern DAWs | Yes |
| Drop a file, get the full-track answer | Pro (paid) | Partial — offline analysis varies by DAW | Yes, free |
| Platform target presets + verdict | Partial (more in Pro) | No | Yes, free |
| CSV / JSON export of the session | Pro (paid) | No | Yes, free |
| Runs without installing anything | No | No | Yes — it's a browser tab |
| Meters live input / another tab's audio | Yes, via your audio device | Routing required | Yes, built in |
Youlean's free meter is genuinely good — it became the default recommendation for a reason, and if you want a meter docked inside your DAW, it's the right call. The browser meter's case is different: nothing to install, nothing gated, and it reads any audio you can get into a tab — including a file you just drag onto it.
BeamTracer reports loudness using the ITU-R BS.1770 and EBU R128 standards. Compare the same complete track with the measurement reset, then choose the workflow that fits your session.
How do you measure a track's loudness in BeamTracer?
Open the Loudness tab, start audio, and read M, S and I as they settle; click ANALYZE FILE when you want the whole-track answer instead of a live pass. Pick your platform in the TARGET card so every reading is judged against it, then export the session as CSV or JSON.
- Open the Loudness meter — the Loudness tab of BeamTracer's Audio Analysis Engine, free, no account
- Start audio: the bundled demo track, a file of your own, or your microphone
- Watch M and S move, and give I at least a chorus to settle
- Click ANALYZE FILE, or drop a track onto the tab, for the whole-file reading in seconds — integrated, true peak, LRA, all of it
- Pick your platform in the TARGET card and the whole meter judges against it — verdict colors, the target band on the graph, the counters
- Press RESET before any live measurement you intend to keep, because integrated loudness accumulates until you do
- Export the session as CSV or JSON, or use COPY for a numeric summary you can paste into notes or a client email
The question-mark button knows where you are
Every reading on the meter is documented — the help button in the corner of the Analysis Engine opens the full Loudness reference for whichever tab you're on, so you never have to leave the meter to look something up.
Next
- What Spotify actually does to your track — normalization up, down, and the true-peak cap nobody models
- How dynamic is your master? PSR, PLR and LRA — the numbers that survive normalization
- Level a multi-person livestream in the browser — the classic two-computer workflow, collapsed into a tab
- Best audio visualizers in 2026 — the rest of what lives behind that same
Akey
Open a broadcast-grade loudness meter in your browser. Free, no install, no account.
OPEN THE LOUDNESS METER →Five numbers, all named. beamtracer.com — see you in the next one.