Audio follows the studio pattern like everything else, with its own category list:

The action goes after the number, and a variation is a two-digit number after that. So a
randomised pool of shotgun fire reads AU_SFX_Wpn_Shotgun_01_Fire_01, …_Fire_02, and so on.

.wav for anything short and triggered. No decode cost, no quality loss on something you will hear
a thousand times. Music and ambience can be .ogg, where the file size matters more than the
handful of milliseconds it costs to stream.

Anything meant to have a position in the world has to be mono. Unity cannot pan a stereo clip
to a point in space, so a stereo footstep plays flatly in both ears no matter where the character
is. Music and ambience stay stereo, because they are not coming from anywhere.

Measured as integrated loudness, with the tolerance either side. Mixing by ear across a session
drifts; mixing to a target does not.

A one-shot longer than this is usually a loop that was not marked as one, or several sounds that
want to be separate clips.