Guide
What LUFS is, and the number to aim for
The short answer
LUFS is a measure of perceived loudness across a whole recording. Podcasts are normally delivered at about −16 LUFS for stereo, and music streaming platforms play back around −14. Matching the target is what makes a recording sit at the same volume as everything else a listener plays.
By Suprabhat Updated How we write these
Why peak volume is the wrong measure
A peak meter shows the single loudest instant in a file. Two recordings can peak at exactly the same point and still sound wildly different in volume, because one is consistently loud and the other is quiet with an occasional bang.
LUFS — Loudness Units relative to Full Scale — measures the average over time, weighted for how sensitive human hearing is at different frequencies. It is the closest thing there is to a number for "how loud does this sound".
This is why "normalize to 100%" in an old audio editor rarely fixes a quiet recording: it scales the file until its loudest peak hits the ceiling, which may raise the overall loudness hardly at all.
The numbers people use
−16 LUFS — the usual target for a stereo podcast, and what most podcast platforms expect.
−14 LUFS — roughly where the major music streaming services normalise playback. Deliver louder than this and they simply turn it down, so mastering it louder gains nothing and costs dynamics.
−23 LUFS — the EBU R128 broadcast standard, used in European television. Much quieter, because broadcast has room for dynamic range.
True peak should stay at or below −1 dBTP in every case, to leave headroom for the distortion that lossy encoding can introduce.
Normalizing versus compressing
Loudness normalization measures the whole file and applies one gain change so the average lands on the target. The relationship between loud and quiet parts is untouched — nothing is squashed.
Compression is a different thing: it reduces the distance between loud and quiet moments, which lets the whole recording sit higher. It is useful for speech recorded at an inconsistent distance from the microphone, but it is an effect, not a measurement.
Normalize Audio does the first: it measures in LUFS, applies the gain, and leaves the dynamics alone. If one speaker is simply too quiet relative to another, Change Volume on the individual file is the more honest fix.
Questions
What LUFS should a podcast be?
About −16 LUFS for stereo, or −19 for mono, with true peaks no higher than −1 dBTP.
Is normalizing the same as turning up the volume?
Not quite. Normalizing measures the perceived loudness of the whole file first and then applies the gain that lands it on a target, rather than scaling by an arbitrary amount.
Will normalizing damage the audio?
No. It is a single gain change, and a good implementation checks that the result does not clip. It is one of the few audio operations that is genuinely safe.