Normalization adjusts the gain of an audio signal to reach a target level. There are two main types:
Peak normalization brings the highest peak in the audio to a target level (e.g., 0 dBFS). Simple but doesn't account for perceived loudness — a quiet track with one loud transient will barely change.
Loudness normalization adjusts the overall level to match a target loudness (e.g., -14 LUFS). This is what streaming platforms use, ensuring all tracks play at a similar perceived volume.
In game audio: Normalization is used to ensure consistent levels across large sound libraries. Many teams normalize SFX to a standard peak level during asset preparation, then use the game engine's volume controls for final mix balance.