Two sounds at the same -6dB peak — sustained music and a single hand-clap — can feel completely different in loudness to human ears, because raw peak level (dBFS) doesn't capture "how loud it feels." LUFS (Loudness Units Full Scale) measures perceived loudness by combining average level over time with how sensitive human hearing is at different frequencies. YouTube, Spotify, and Netflix normalize everything to roughly -14 to -16 LUFS so you don't have to reach for the volume knob between videos.
When several people build UI sounds from different raw waveforms (one loud sawtooth, one quiet sine), the finished set often feels wildly inconsistent in loudness once you hear it on a real device. Always run a normalization pass before shipping — measure every sound asset's LUFS and bring it to a target (mobile apps typically aim for -18 to -23 LUFS). Matching peaks alone isn't enough — a sound can still feel uniquely loud even at the same peak level.
Used to unify volume across content on streaming platforms, keep a consistent felt loudness across an app's notification sound set, and fix the "loudness war" where ads feel much louder than the program.
When to use
Run it before shipping a set of sound assets together. Match perceived loudness by LUFS, not just peak level.