Introduction
Upload a song to a streaming service or an episode to a podcast app, and it will usually be played back at a standard loudness. If your file is much quieter, it sounds weak next to everything else; if it is much louder, the platform turns it down and the extra level buys you nothing.
The fix is loudness normalization: measure how loud the whole file sounds, in LUFS, and move it to the target the platform expects. This guide explains the targets, then walks through doing it in ToolBuddy's Audio Normalizer, which runs entirely in your browser.
LUFS, peak and why they differ
Peak level is the single loudest sample in a file, measured in dBFS. It tells you how much headroom is left before clipping, but almost nothing about how loud the file sounds: a quiet recording with one sharp click can peak at full scale.
Loudness, measured in LUFS, describes how loud the whole file sounds to a listener. It is measured with ITU-R BS.1770, the method EBU R128 and the streaming services use: frequencies the ear is most sensitive to count more, the file is measured in 400 ms blocks, and silence and very quiet passages are left out so they don't drag the reading down.
Two files at the same LUFS sound about equally loud. That is why platforms level by LUFS, and why you should too.
- Peak (dBFS): the loudest sample; use it to protect against clipping.
- Loudness (LUFS): how loud the file sounds overall; use it to match a platform.
- 1 LU is 1 dB, so -14 LUFS is 2 dB louder than -16 LUFS.
Which target should you use?
Each platform publishes a reference loudness. Matching it means your audio plays at the intended level without being turned up or down. The table lists common targets; check the current guidance for the platform you deliver to, as they occasionally change.
| Destination | Target | Preset in the normalizer |
|---|---|---|
| Spotify, YouTube and most music streaming | About -14 LUFS | Streaming |
| Apple Podcasts and most podcast hosts | About -16 LUFS | Podcast |
| EBU R128 television and radio | -23 LUFS | Broadcast |
| Audiobook and narration platforms | Varies, often about -18 to -20 LUFS | Custom |
| Game audio, internal review | Whatever your project uses | Custom |
Load and measure the file
Open the Audio Normalizer and drop a file on the selector, or tap Select audio file. It accepts MP3, WAV, M4A, AAC, OGG/Opus, FLAC and WebM audio up to 500 MB (100 MB on phones and tablets). The file is decoded on your device and nothing is uploaded.
As soon as it loads, the Control Panel shows the Source loudness in LUFS and the Source peak in dBFS. That is your starting point: a loud, heavily mastered song might read -8 LUFS, a quiet phone recording -30 LUFS.
Choose loudness or peak
Mode is set to Loudness (LUFS) by default. Choose a target: Streaming (-14 LUFS), Podcast (-16 LUFS), Broadcast (-23 LUFS) or Custom, which adds a slider from -30 to -8 LUFS. The Gain readout updates immediately and shows how much the file will be raised or lowered.
Switch Mode to Peak when you only need headroom, for example before sending a file to someone who will mix it. Peak mode moves the loudest sample to the target peak, from -6 to 0 dBFS.
Set the peak ceiling so nothing clips
Raising a quiet file to -14 LUFS can push its loudest moments past full scale. The peak ceiling, -1 dB by default, is the highest the peak is allowed to go. The normalizer applies one clean gain to the whole file and stops that gain at the ceiling.
If the target needs more gain than the ceiling allows, the gain stops at the ceiling, and after you normalize the Result loudness shows how loud the file got. You can lower the target, raise the ceiling towards 0 dB, or accept a result slightly under the target, which platforms will simply play a little quieter. The tool does not compress or limit, so the dynamics of the recording are not changed.
Normalize, compare and export
The output starts in your file's own format and quality. Change it if the file is headed somewhere specific: WAV or FLAC to keep editing, MP3 or M4A to upload or share.
Select Normalize audio. The Control Panel adds the Result loudness and Result peak, and the preview switches to the normalized file. Use the Original and Normalized buttons to compare; the player stays at the same moment, so you hear the change on the same words or notes. When it sounds right, select Download.
| Output | Best for | Note |
|---|---|---|
| WAV | Further editing, archiving | Lossless; largest files |
| FLAC | Archiving at a smaller size | Lossless; not every app plays it |
| MP3 | Uploading and sharing | Keep 128 kbps or higher for music |
| M4A / AAC | Apple devices, small files | Depends on browser support |
| OGG | Web and games | Depends on browser support |
Common jobs, step by step
A few normalizations come up again and again.
- Podcast episode: Loudness, Podcast preset, ceiling -1 dB, export MP3 at 128 kbps or higher.
- Song for streaming: Loudness, Streaming preset, ceiling -1 dB, export WAV or a high-bitrate MP3.
- Broadcast spot: Loudness, Broadcast preset, keep the ceiling at -1 dB or lower if your broadcaster asks for it.
- Quiet voice note: Loudness, Podcast or Custom around -18 LUFS, then compare to make sure the noise floor is not too loud.
- Stems for a mix: Peak mode at -6 dBFS to leave headroom for the engineer.
What normalization does not do
Normalization changes the overall level of a file. It does not make quiet and loud passages closer together; that is compression, and a very dynamic recording can be capped by the ceiling before it reaches -14 LUFS. It also does not remove noise: raising a quiet recording raises its background hiss too.
For those jobs, use the Audio Editor or the Noise Remover first, then normalize. To bring several clips to the same level before joining them, normalize each one to the same target, then use the Audio Joiner.
Written by
Ankit Kumar
Senior Software Engineer & Founder at ToolBuddy
Engineering browser-first utilities with zero upload requirements using WebAssembly, Web Audio, and Canvas technologies.