Why does my normalized file still sound quiet on a phone?

Small speakers and aggressive platform processing color perception. LUFS normalization is still the right target — compare on the same device before blaming the number.

Can normalization fix bad recording levels?

It fixes level, not signal-to-noise. A quiet recording normalized to -16 LUFS brings its noise floor up too; denoise first, normalize second.

Parameters 6 min read

Audio Too Quiet or Loud: Gain, Loudness and Batch Normalization

Two files can peak at the same level and one still sounds quiet — peaks are instantaneous, perceived loudness is averaged. Loudness normalization (LUFS) fixes what peak gain cannot.

Three tools, three jobs

Gain (+/- dB): flat shift, use when you know the offset. Clipping risk if pushed too far.
Peak normalize: sets the highest sample to a target (e.g. -1 dBTP). Does not equalize perceived loudness.
Loudness normalize (EBU R128 / LUFS): measures integrated loudness and corrects to target — the modern default for distribution.

Targets that work everywhere

Podcasts: -16 LUFS stereo (-19 mono) per the common platform guidance.
YouTube: -14 LUFS.
Broadcast: -23 LUFS (EBU R128 standard).
Keep true peak under -1 dBTP to survive lossy encoding without clipping.