Audio Normalizer

Listeners notice level jumps more than almost any other problem: an episode that is much quieter than the one before, or an intro that blasts after a calm interview. Platforms normalize playback to their reference level, so a track mastered far louder is simply turned down — and loses punch — while a quiet one may be left quiet. Delivering at the target keeps your audio consistent with everything around it.

Each platform plays content at a reference loudness. Delivering close to it means your audio is played as you intended.

The measurement follows ITU-R BS.1770. The sound is first filtered with K-weighting, which reduces the lowest frequencies and boosts the highs slightly to match how we hear. The energy is then measured in overlapping 400-millisecond blocks. Blocks quieter than −70 LUFS, and blocks more than 10 LU below the average, are left out (gating), so silences and pauses don't drag the result down. The result is the integrated loudness of the whole file. The peak is the largest sample value; true-peak measurement between samples is not included.

Raising a quiet recording to a loud target can push its peaks above full scale. The limiter, which is always on here, smoothly bends those peaks below −1 dBFS instead of clipping them, and the result message says how many samples it touched. For speech and most music this is inaudible. If a dynamic recording needs a lot of limiting to reach −14 LUFS, a gentler target such as −16 LUFS will sound more natural.

Normalizing uses a measurement to reach a standard level automatically, which is ideal for podcasts, audiobooks, voice-overs and playlists. To change the level by a specific amount instead, use the Audio Volume Changer. Lossy formats are re-encoded, so choose a high bitrate or WAV to keep quality, and keep your original file.