Audio Normalizer

Simple volume boosting (or peak normalization) raises all audio by the same number of decibels so the loudest peak hits a target level. It doesn't account for how loud the audio actually sounds to human ears. A heavily compressed track with lots of loud material will sound louder at the same peak level than a dynamic track with quiet passages and occasional loud peaks.

EBU R128 loudnorm, used by this tool, measures integrated loudness — a weighted average of loudness across the entire file, measured in LUFS (Loudness Units Full Scale). This matches how human hearing perceives loudness much more accurately. Two files normalized to -14 LUFS via EBU R128 will sound approximately equally loud on playback, regardless of their dynamic range or peak levels.

LUFS (Loudness Units Full Scale) is a measure of perceived loudness. The more negative the number, the quieter the audio. The industry has converged on two main targets:

-14 LUFS (Streaming): Spotify targets -14 LUFS before making playback adjustments. Apple Music and YouTube target similar levels. For music, this is the standard. Content mastered significantly louder than -14 LUFS will be turned down by the platform; content quieter will be turned up. Delivering at -14 LUFS means what you hear in your browser is what listeners hear on Spotify.

-16 LUFS (Podcast): The podcast industry converges around -16 LUFS, with Apple Podcasts and Spotify for Podcasters using this as their guideline. It's 2 LU quieter than streaming music, which gives voice content slightly more headroom for dynamic range — a louder laugh or emphasis doesn't clip as easily at -16 LUFS as at -14 LUFS.

Alongside the integrated loudness target, the loudnorm filter also applies a true peak ceiling at -1 dBFS (TP=-1). True peak accounts for inter-sample peaks that occur during digital-to-analog conversion — peaks that can cause clipping even when the digital waveform looks safe at -0.1 dBFS. Setting the ceiling at -1 dBFS provides 1 dB of headroom for these inter-sample peaks, which is the standard recommendation for streaming delivery and broadcast.

This is the same ceiling used by most mastering engineers when preparing audio for streaming. You don't need to worry about it — it's applied automatically based on the preset you choose.

Loudness normalization with server-side tools means uploading your audio to a remote server, waiting for processing, and downloading the result — with your audio passing through infrastructure you don't control. This tool uses FFmpeg compiled to WebAssembly. All processing happens in your browser's memory. Nothing is uploaded. The WASM bundle is cached after the first visit. For audio that you haven't released publicly (unreleased music, client deliverables, confidential recordings), the privacy difference is real and significant.