Audio Volume Normalizer (Peak)

Automatically bring the loudest peak of a WAV or MP3 file to a precise target level (in dBFS), to even out the perceived volume of several files or avoid audio that's too quiet. This is peak normalization, not perceived loudness — see the FAQ below for the difference.

Example:
0 dBFS = maximum volume without clipping. -1 dBFS is a common value to keep a small safety margin.
16-bit PCM WAV or MP3 — 100 MB max per file

How It Works

The tool analyzes the loudest sample (the peak) of your audio file, then applies a constant gain to the entire signal so that this peak reaches exactly the target level you chose, in dBFS (decibels relative to the maximum possible digital value). It's a simple, fast, and predictable operation.

Peak normalization or loudness (LUFS) normalization: what's the difference?

This tool performs peak normalization: it looks only at the loudest sample and adjusts it to the requested level, regardless of the average perceived volume over the whole track. Streaming platforms (Spotify, YouTube...) actually use perceived loudness normalization (LUFS / EBU R128), a far more complex measurement that accounts for the human ear's sensitivity to different frequencies and for how long the sound lasts. Two files normalized to the same peak can therefore have a different perceived volume. This tool doesn't perform that LUFS calculation — don't present it as such.

What target level should I choose?

-1 dBFS is a common and safe choice: it leaves a little headroom to avoid any clipping during a later conversion. Use 0 dBFS for maximum volume without distortion, or a more negative value (-6, -12...) if you deliberately want to keep headroom, for example before mixing.

What happens if my file is already at the target level, or is silent?

If the current peak already matches the target level, the file comes out virtually unchanged. If the file is completely silent (no non-zero samples), there is no peak on which to apply gain: an error message appears for that file instead of a normalized result.

100% local processing: reading, any MP3 decoding, and normalization happen entirely in your browser via WebAssembly — your audio is never sent to a server.