All voice audio (dialogs, words, etc), should be MONO, 44.1KHz files and should be well trimmed at beginning (at zero crossing!)
Level shoudl be normalized with a peak reference at -6db
In the Unity project they are saved in MP3 format, with middle quality VBR (Variable Bit Rate).
To easy the voice audio normalization we have a macro script for Audacity: you can find it here: Antura Normalize Vocabulary - Audacity macro.txt
because we have lots of voice audio files and don’t want to waste app space, and voice recordings just need mono and medium compression.
Audacity is a perfect free multiplatform tool to edit audio files.
on macOS we use the free utility xAct to batch convert audio files.