Audio Loudness Normalizer
Measure how loud your audio really is, bring it to the level your platform expects, and see the result measured from the file you downloaded. Uses EBU R128, the standard streaming services use.
Normalize Audio
Add your files. Recommended settings are ready, so you can process in one click.
PDF, Word, images, audio, video, archives and more
Up to 100 files · 250 MB · No account- 1Add files
- 2Choose an action
- 3Download or keep going
What happens to your file.
Loudness is measured twice. Your browser measures the file you add using the ITU-R BS.1770 method — K-weighting, 400 ms blocks, and the two-stage gate that stops pauses between words from dragging the figure down — so you see what you have before anything is sent anywhere. The normalisation itself runs on our server with FFmpeg: a first pass measures the file, and a second applies a single constant gain calculated from that measurement. When the file comes back, your browser measures it again, and that second measurement is what the report shows.
Where it stops: This tool sends your file to our server, because the processing needs FFmpeg. The file is written to a temporary folder for the request and deleted as soon as it finishes; it is not kept or used for anything else. Normalisation adjusts level, not quality — background hiss gets louder along with the voice, so clean up noise first if the recording is poor. If reaching the target would push peaks past your ceiling, the result lands slightly below the target rather than clipping, and the page says so beforehand.
Which target should you choose?
| Where it is going | Target |
|---|---|
| Podcast (Apple, Spotify) | -16 LUFS, peak -1 dB |
| Music streaming (Spotify, Apple Music, YouTube) | -14 LUFS, peak -1 dB |
| European broadcast (EBU R128) | -23 LUFS, peak -1 dB |
| US broadcast (ATSC A/85) | -24 LUFS, peak -2 dB |
| Audiobook (ACX) | -18 LUFS, peak -3 dB |
Why two passes matter
Running a loudness filter in one pass makes it adaptive: it adjusts the gain as the file plays, which is audible as pumping on speech and does not actually land on the target. Measuring first and then applying the result as a single constant gain is what "normalised to -16 LUFS" is supposed to mean — the whole file moves by one amount, the relationship between loud and quiet passages is untouched, and the finished file measures at the target.
Loudness or peak?
- Loudness (LUFS) matches how loud the audio sounds to a listener, measured across the whole file. This is what every platform means by normalised, and what you almost always want.
- Peak only scales the single loudest sample to your ceiling. It changes nothing about perceived loudness — a quiet recording with one loud cough stays quiet — and is mainly useful as a safety step before mastering.
What people use it for.
Everything you need, without a paywall.
Common questions.
How is this different from turning the volume up?
Raising volume scales everything, so a single loud peak limits how far you can go. Loudness normalisation measures perceived average level over the whole file, ignoring silence, and moves that to your target — which is what listeners and platforms actually notice.
What target should I use for a podcast?
-16 LUFS with a -1 dB peak ceiling is the usual target for stereo podcast distribution. Music streaming platforms sit nearer -14, and European broadcast at -23.
Is my audio uploaded?
The measurement happens in your browser, so you can see what you have without sending anything. Normalising does send the file to our server, because it needs FFmpeg; it is written to a temporary folder for that request and deleted as soon as the job finishes.
How do I know it actually hit the target?
The file that comes back is measured again in your browser, and the report shows that measurement next to the target. Those figures are read from the file you downloaded, not copied from the request.
Will it remove background noise?
No, and raising the level makes existing noise more audible. Reduce noise first if the recording has hiss or hum.
Will my file stay the same format?
Yes, by default, and the sample rate is preserved too. You can choose MP3, WAV, FLAC or M4A instead if you need a specific format.
What if my audio is already louder than the target?
It is turned down. Normalising works in both directions — the point is to land on the target, not to make things louder.
What does loudness range tell me?
How much the level varies between quiet and loud passages. A tightly controlled broadcast mix might be 3 to 5 LU; a film score can be 15 or more. A very high range on speech usually means one speaker is much quieter than another.
You might also need.
Reduce audio size by re-encoding to MP3 or AAC at a selected bitrate while preserving normal playback compatibility.
Raise or lower audio volume by a percentage and export a new MP3 or WAV file.
Draw the waveform of a recording as a PNG or SVG in your colours, or export its peaks as JSON for a web player.
Convert M4A audio to MP3 for broad compatibility across players and devices.