How to make audio louder without distortion (and what LUFS are)

A recording that's too quiet forces listeners to turn their volume all the way up, and then the next audio blasts their ears. Turning up the volume sounds like the simplest thing in the world, but done carelessly it produces digital audio's ugliest flaw: clipping, that harsh crackle that can't be fixed afterwards. Three concepts are enough to avoid it.

Three concepts worth understanding

• The peak is the single loudest instant of the signal. In digital audio it's measured in dBFS, and the absolute maximum is 0 dBFS: nothing in the file can be louder than that.
• Clipping happens when you amplify so much that peaks would go past 0 dBFS. Since they can't, they get cut off and the wave is distorted. It sounds like crackling or harshness.
• Loudness (LUFS) measures how loud the audio sounds overall, the way the ear perceives it, not just its highest point. It's the reference streaming platforms, podcasts and broadcasters use.

The key is that two clips with the same peak can sound very different. A voice with one loud moment (a laugh, a clap) has a high peak but sounds quiet overall. Boost it crudely and the clap clips before the voice becomes clearly audible. That's why loudness normalization with peak control works better.

Boost or normalize

Which level to use?

Many music platforms normalize on their own: upload a very loud song and they turn it down so it doesn't stand out from the rest. That's why there's no point 'winning the loudness war' by crushing a mix: you just lose dynamics. For speech, -16 LUFS is a comfortable, widely used target.

The right order

1. Remove the noise first with the noise remover. Boost the volume first and you boost the background noise too.
2. Equalize if needed, with the equalizer: a voice with too much low end will seem quieter than it is.
3. Normalize last. It's the final step before exporting, so nothing changes the level afterwards.

Matching several files

If you have several episodes, audiobook chapters or songs from different sources, normalize each one: they'll all end up at -16 LUFS and play at a similar volume. If you also want them in a single file, join them afterwards with the audio joiner.

How to tell if audio is clipped

Listen on headphones to the loudest moments: drum hits, laughs, plosive consonants like 'p' or 'b'. If they sound harsh, broken or with a crackle that isn't in the rest, there's clipping. In the waveform it shows as peaks 'cut flat' against the top and bottom edges. If it happens after boosting, lower the percentage or switch to Normalize mode; if it was already in the original, the fix is to record again with less gain.

What boosting volume can't fix

• Audio that's already clipped. If the original crackles, raising or lowering the volume won't remove it.
• A distant voice. If it was recorded far from the microphone, it will be louder but still sound distant and echoey.
• Missing bass. If what you miss is body, try Bass Boost in moderation.

FAQ

Can normalizing ever lower the volume?

Yes. If the file already sounds louder than -16 LUFS, normalization brings it down to that level. That's what lets several files sound the same.

Is quality lost?

Changing the volume doesn't degrade audio as long as it doesn't clip. The only possible loss comes from the export format: if you save as MP3, choose 192 kbps or more.