Loudness Range (LRA) Explained: The Number Behind "Easy to Listen To"

Two episodes measure the same -16 LUFS. One feels calm and clear. The other makes you ride the volume knob — whispers you cannot hear, then a laugh that hurts. The difference is loudness range: how much the loudness moves inside the file. This guide explains LRA, what healthy values look like, and how to fix a show that swings too much.

LRA appears in every LoudNorm report next to integrated loudness and true peak, because the three numbers together describe a file. The LUFS explainer covers the other two.

What LRA Measures

Loudness range is measured in LU (loudness units). The meter looks at the loudness moment by moment, throws away the very quietest and loudest extremes, and reports the spread of what remains. A file that sits at one steady level has a small LRA. A file that moves between whispers and shouts has a large one.

The method, from EBU Tech 3342, works like this: the meter takes short-term loudness readings through the whole file, builds a picture of how the readings are spread out, and then reports the distance between the 10th and the 95th percentile — in plain words, it ignores the softest tenth (room tone, pauses, fade-outs) and the loudest sliver (the single biggest laugh), and measures the honest middle where the programme actually lives. That is why LRA describes typical movement, not freak events, and why it pairs so well with the integrated number: integrated LUFS tells you where the file sits, LRA tells you how much it wanders around that seat.

Why LRA Matters for Podcasts

Podcasts are heard in cars, kitchens and trains — noisy places where quiet passages vanish. A moderate LRA (roughly 5–10 LU) keeps every word audible without flattening the life out of the conversation. A very large LRA usually means the real problem is balance: one guest much quieter than the other, or a microphone too far away. A very small LRA means heavy compression, which sounds loud but fatiguing across a 40-minute episode.

Think about the listening environments honestly. A car at highway speed has a noise floor around conversation level itself. Any speech that dips 10 LU under your episode's average is, for that listener, simply gone — and a listener who misses every third sentence does not blame your LRA; they blame your show and unsubscribe quietly. At the other extreme, a caller listening at night next to a sleeping partner needs the loud moments contained. LRA is the dial between "alive" and "audible," and the middle of the dial is where speech belongs.

Two Episodes, Same LUFS, Different Shows

Here is the comparison this whole page exists for. Both episodes measure exactly -16.0 LUFS integrated:

Episode B is not badly recorded — it is a lively interview where one guest leans in and out of a microphone. But its quiet guest spends long stretches 11 LU under Episode A's quiet passages, and in a car those stretches disappear. Normalizing Episode B to -16 was correct and necessary; it was never going to be sufficient. LRA is how you know that before your listeners tell you.

What LRA Should Your Show Aim For?

Use the content, not a rulebook:

• Solo speech and narration: 4–8 LU. One voice, one distance from the mic — the range should naturally be small. A reading above 10 LU in a solo show usually means mic drift or wandering energy.
• Two-host conversation: 5–10 LU. Some swing is the sound of two real people; the goal is that neither voice ever gets lost.
• Interviews with remote guests: 6–12 LU, and treat anything above that as a balance problem to fix at the source.
• Storytelling with music and scenes: 8–15 LU can be right by design — drama needs contrast — but check the quiet scenes on earbuds in a noisy room before you commit.

These are podcast working ranges, not standards. No platform rejects a file for its LRA. The number is a diagnostic: it tells you where to spend mix effort, and whether the effort worked.

How to Fix a Swinging Episode

Normalization alone does not change LRA — it moves the whole file up or down together. To narrow the range you need dynamics work before the loudness step:

1. Level the voices first. Bring the quiet guest up and the loud one down, clip by clip or with a leveling pass.
2. Compress gently. A light compressor (2:1, slow attack) tames the biggest swings without pumping.
3. Then normalize. With the range under control, the two-pass loudness step lands the file on target — the workflow in the podcast guide.

If the episode is simply quiet overall rather than uneven, the quiet recording guide is the better starting point.

A note on order: LRA work always comes before loudness normalization, and loudness normalization is always last. Compress after normalizing and you change the integrated number you just certified. The sequence — balance, dynamics, then loudness — is the whole recipe; each step assumes the previous one is finished.

When a Small LRA Is the Problem

Not every LRA complaint means "too wide." An episode crushed to 2–3 LU measures beautifully consistent and sounds exhausting: every breath, whisper and shout arrives at the same insistent level, and the ear gets no rest for 40 minutes. This usually comes from stacking loudness tools — a compressor, then a maximizer, then normalization with a busy limiter — each one reasonable, the pile-up fatiguing.

The tell is listener language: nobody writes "your LRA is too small." They write "the show feels intense," "I can only listen in short bursts," or they just stop finishing episodes. If your LRA report reads under 4 LU for speech, back off one stage of processing and listen again. Dynamics are not the enemy of loudness; they are the reason loudness can be comfortable.

A Worked Fix: From 16 LU to 8 LU

A two-person interview comes in at -17.5 LUFS with an LRA of 16 LU. The host is steady; the guest, on a laptop microphone across the room, drifts. The repair, in order:

1. Level the guest's clips. The guest's track comes up about 4 dB overall, with the worst drifts — where they leaned back — raised a further 3 dB by hand. LRA after this pass: about 11 LU.
2. Gentle bus compression. A 2:1 compressor with a slow attack catches the guest's sudden emphatic moments and the host's laughs. LRA: about 8 LU.
3. Normalize. Two-pass to -16 LUFS with the -1 dBTP ceiling. Final report: -16.1 LUFS, true peak -1.0 dBTP, LRA 8 LU.

Nothing about the conversation changed. The guest still sounds like the guest. But no sentence now falls more than a few LU under the show's average, and the car listener hears the whole interview. Total extra work over a straight normalization: perhaps fifteen minutes — once you know the order, and the order is the guide above.

Auditing Your Back Catalogue in Five Minutes Per Episode

You do not need to re-listen to your archive to find its LRA problems. Run the catalogue through measurement — a batch on a paid plan returns one report for every file — and sort by LRA. The episodes above 12 LU are your volume-knob episodes; the ones under 4 LU are your fatigue episodes. Fix the worst offenders first, in the repair order, and re-check their reports. Most catalogues have a handful of bad episodes and a long tail of fine ones; the report tells you which is which instead of your memory guessing.

LRA in Broadcast

Broadcasters set LRA expectations per programme type under EBU R128 — a film may roam widely, a news bulletin may not. The delivery side is covered in the EBU R128 guide.

Broadcast practice is a useful sanity check for podcasters. News and speech programming on European television typically ships with modest ranges not far from the podcast working bands above, because the problem is identical: speech must stay intelligible in real rooms. Drama and film go wider because the audience is seated, the room is quiet, and the contrast is the art. Your podcast is speech in noisy rooms — program like the news, not like the cinema.

Common Mistakes

Trying to fix range with normalization. Turning the file up makes the quiet parts louder and the loud parts louder by the same amount. The gap — the thing the listener fights — is untouched.

Compressing the master to fix one bad segment. If minutes 12–15 are the problem, fix minutes 12–15. Whole-file compression to rescue one stretch taxes the 35 good minutes for the 3 bad ones.

Chasing an LRA target during recording. LRA is a finishing diagnostic, not a performance note. Hosts who perform "consistently" because a meter is watching sound like newsreaders having a hostage situation. Record naturally; shape afterwards.

Reading LRA without LUFS, or LUFS without LRA. Either number alone misleads. -16 LUFS with 17 LU is a workout; -22 LUFS with 5 LU is a pleasant show recorded too quietly — one needs mix work, the other just needs gain, and the two reports together tell you which is which.

Frequently Asked Questions

Does any platform require a specific LRA? No podcast or streaming platform grades LRA. Broadcasters describe genre expectations for it, but for podcasters it is guidance for your own quality control — arguably the most useful number nobody makes you look at.

Can normalization change my LRA? Plain gain cannot. Limiting can, slightly: a normalizer holding peaks with a limiter will shave the loudest moments, which can trim the top of the range. On typical speech the effect is small — a fraction of a LU — and the report shows the result either way.

What is a good LRA for an audiobook or long narration? Similar to solo podcasting: roughly 4–8 LU keeps narration steady across chapters and listening sessions. Audiobook listeners binge for hours, so err toward the calm end.

My LRA is fine but one moment still blasts. LRA is a spread of typical values; a single freak event — a dropped microphone, a shout into the mic — lives in the extremes the method throws away. Find those by waveform or by ear, fix them individually, and let LRA judge the programme as a whole.

Is a bigger LRA ever a selling point? For speech, no listener has ever praised a show for making them work harder — range is a means to natural delivery, not a feature. For drama and music it can be artistry. Know which product you are making, and let that decide the number you are happy with.

Should music podcasts watch LRA too? Yes, but read it like a musician: music legitimately roams wider than speech. Watch instead for the talk segments — an intro recorded at a very different level from the tracks is a balance problem that LRA across the whole file will expose.

References

• https://tech.ebu.ch/publications/r128 — EBU R128 and its loudness range measurement (EBU Tech 3342).
• https://www.itu.int/rec/R-REC-BS.1770 — ITU-R BS.1770, the base loudness measurement.
• https://www.apple.com/apple-podcasts/ — Apple Podcasts, whose ecosystem made consistent episode loudness a listener expectation.