Audio Tools New Tool

Free Online Audio Normalizer With Real LUFS Measurement

Normalize audio loudness free without uploading anything. It uses the real broadcast measurement, filter and both gates included, so a batch actually ends up level with itself, and a true peak ceiling stops the level change from clipping the file.

Free Forever Nothing Uploaded Real LUFS, Gated True Peak Ceiling

Free Online Audio Normalizer With Real LUFS Measurement

Advertisement Slot (Top Banner) Google AdSense Unit • Responsive Banner

Drop audio files here to even out their loudness Bring a whole folder to the same measured loudness, the way streaming services do it. Select Audio Files MP3, WAV, M4A, OGG, FLAC and more • best under 25 minutes • nothing is uploaded

The shape of your sound will appear here.

0:00.000 / 0:00

What to match

The target

Loudness target

Never let the true peak go above

What this file measures now

Loudnessnot measured yet

True peak-

Change needed-

Will land at-

Measuring loudness properly takes a moment, so it is not done automatically for every file you drop in. Running the job measures each file anyway.

Output

Format

Bitrate

Sample rate

Channels

WAV detail Only used when the output is WAV. The compressed formats have their own.

Encoder

File name

.mp3

Buy Us A Coffee

Enjoying WizTools123? Help keep our server infrastructure 100% free and open for everyone.

Buy Us A Coffee

Sponsored Content (Below Tool) Google AdSense Placement

How to Normalize Audio

Pick a target, see what each file measures now, run it.

Drop the files in One file or a whole album. Each one is measured on its own, which is the only way a batch can end up level with itself.

Choose the target Minus fourteen for streaming, minus sixteen for podcasts, minus twenty three for European broadcast. If you are not sure, minus fourteen is the one most people want.

Press measure to see where a file sits This is optional. It shows the current loudness, the true peak, and exactly how much the file will move, before anything is written.

Run it and take the files Each file is measured, moved, and checked against the peak ceiling. Anything that had to be held back is reported rather than hidden.

Peak and Loudness Are Not the Same Thing

The difference that makes one of these actually work.

Peak and loudness are completely different measurements. Peak is the single highest sample in the file. Loudness is how loud the whole thing sounds to a person, measured the way broadcasters measure it. Two files can have identical peaks and be twice as loud as each other, which is exactly why peak normalisation never makes a playlist sit level.

The loudness figure ignores silence, on purpose. The standard measures the file in short blocks, throws away the ones that are near silence, and then averages what is left. Without that, a track with long quiet passages would read far quieter than it sounds, and would then be pushed up too far.

Raising the loudness can push the peaks through the ceiling. This page measures the true peak afterwards and pulls the level back down if it has to, and tells you when it did. That means the result may land slightly below the target you asked for, which is the correct behaviour rather than a fault.

This does not compress or limit the sound. The whole file is moved by one amount. Bringing a very dynamic recording up to a loud streaming target without a limiter is not possible, and this page will tell you when the target is out of reach rather than clipping its way there.

The work happens on your own machine. Your file is opened by this page, not sent anywhere. That is why there is no upload bar, no queue and no waiting for a server. It is also why a long recording leans on your own memory: nothing is being done for you somewhere else.

Most of what this page does needs no download at all. Your browser already has a full audio engine in it, and that is what does the work. A thirty megabyte encoder is only fetched if you ask for MP3, M4A, OGG or FLAC at the end. Ask for WAV and nothing is downloaded, ever.

Length matters more than file size here. A browser opens sound at full detail, which works out at about twenty one megabytes of memory for every minute of stereo. A three minute song is nothing. A two hour recording is a different matter, and this page tells you where you stand the moment you pick a file.

MP3, M4A, OGG and Opus throw information away, every time. Opening one and saving it again is a second pass of that, so the result is always slightly worse than the original even if you changed nothing. Where it matters, take WAV or FLAC out of this page instead.

Key Features & Capabilities

What this tool does, and what it deliberately does not.

Nothing uploaded A whole album measured and normalised on your own machine, with nothing sent anywhere.

Real LUFS, gate and all The broadcast measurement with its weighting filter and both gates, not an average dressed up as one.

True peak ceiling Measured between the samples, where cheap converters actually clip, and enforced after the level change.

The standard targets Spotify, YouTube, Apple, podcast and broadcast figures, each labelled with what it is for.

A whole folder at once Every file measured separately, which is the only way a batch ends up level with itself.

Says when it could not reach If the peak ceiling stops a file getting to the target, you are told by how much rather than left guessing.

About the Bulk Audio Normalizer

Normalising audio is one of those jobs where the common tool does the wrong thing convincingly. Most normalisers find the single loudest sample in a file and scale everything so that sample lands at a chosen ceiling. It is fast, it is exact, and it almost never produces the result people wanted, because the loudest sample tells you nothing about how loud a recording sounds. A heavily compressed pop track and a quiet acoustic recording can have identical peaks and differ by ten decibels in apparent loudness.

What actually corresponds to loudness is a measurement the broadcast world settled on years ago and the streaming services then adopted. The sound is first passed through a filter that approximates the way an ear weights different frequencies. It is then measured in short blocks rather than as a whole, and the blocks that are near silence are thrown away before anything is averaged, so long quiet passages do not drag the figure down. The result is a number in LUFS, and two files with the same LUFS really do sit at the same level.

This page implements that measurement properly, including both of the gates the standard specifies, which is why its figures will not agree with a simple average and should not be expected to. It then moves the file by the difference between what it measures and what you asked for. That is a single gain change across the whole file: nothing is compressed, and the relationship between the loud and quiet parts is untouched.

The complication is peaks. Raising a file to a loud target often pushes its peaks above the ceiling, and a peak that goes above the ceiling is clipped, which is audible. So after the level change the true peak is measured again, at four times the sample rate, because the wave between two samples can rise higher than either of them and that is where converters actually distort. If the peak would exceed the ceiling, the level is pulled back and the page says so. That means a very dynamic recording may land a decibel or two below the target you asked for, which is correct: the alternative is a distorted file.

What this page will not do is limit or compress to force the target. That is a deliberate omission, not a missing feature. Squashing dynamics to reach minus fourteen is a mixing decision with consequences, and burying it inside a button called normalise is how recordings quietly get ruined.

Frequently Asked Questions

LUFS targets, peak normalisation, streaming and batches.

Peak normalising finds the single loudest sample and moves the whole file so that sample sits at a chosen level. LUFS normalising measures how loud the file actually sounds and moves it so that measurement hits a target. Only the second one makes different tracks sit at the same apparent level, which is what people almost always want.

Minus fourteen is what Spotify, YouTube and Apple Music normalise to, so it is the sensible default for music. Podcasts usually aim at minus sixteen. European broadcast is minus twenty three and American broadcast is minus twenty four. If your material is going somewhere with a published figure, use theirs.

Because getting to minus fourteen would have pushed its peaks above the ceiling. Rather than clipping, the page holds the level back and tells you. That happens with very dynamic material, and the proper fix is a limiter, which this page deliberately does not include.

Peak is the highest sample actually stored in the file. When those samples are turned back into a continuous wave for playback, the wave between two samples can go higher than either of them. That higher value is the true peak, and it is where cheap digital to analogue converters distort. It is measured here by oversampling four times, which is what the standard specifies.

No. It applies one level change to the entire file, so the relationship between the loud and quiet parts is exactly as it was. Nothing is squashed. That is also why a very dynamic recording cannot always reach a loud target.

Yes, and that is the main reason this page handles batches. Each track is measured on its own and moved to the same target, which is what makes them sit level with each other. Note that this normalises each track individually rather than preserving the relative levels an album mix intended.