Compression
Automatically turning down the loudest moments so you can raise everything else, making a voice sit at an even, present level.
See it
What it is
A compressor watches the signal and turns down anything above a set threshold, by an amount set by the ratio (3:1 means 3 dB over the line comes out as 1 dB over). That alone makes things quieter, so you add makeup gain to bring the whole thing back up. Net effect: the loud bits came down, everything else went up, and the performance now sits at a steady, present level. Attack and release control how fast it grabs and lets go.
Reach for it on anything with a human in it: voiceover, dialogue, interviews, vocals. A sane starting point for spoken word is a ratio around 3:1 with the threshold set so the loudest phrases show 3 to 6 dB of gain reduction, attack near 10 ms so consonants keep their snap, release near 100 ms.
Gotcha: this has nothing to do with MP3 'compression', which is file size. And a compressor is an amplifier for everything you did not want: push it hard and the room tone, mouth clicks, chair creaks, and hiss all rise into view between words. Two gentle compressors in series beat one brutal one.
Ask AI for it
Apply gentle dynamic range compression to this voiceover so it sits at an even level: ratio 3:1, threshold set so the loudest phrases get 3 to 6 dB of gain reduction, attack around 10 ms so consonants stay crisp, release around 100 ms, then makeup gain to restore the original loudness. Keep it transparent: no pumping, no breathing between words, and do not let the background room noise rise audibly in the pauses.