A narrator is the person performing an audio story. In most fiction the narrator serves the text; in audio erotica the relationship is closer to the other way round, and listeners routinely pick what to play by who is reading it.
How the apps handle credit
Inconsistently. Some services name the performer on every story and let you browse by them, which is what listeners want. Others keep narrators anonymous or use stage names only, and a few do not credit them at all. Where you cannot follow a voice, the catalogue is harder to use than its size suggests.
Hired, licensed or uploaded
Three models. A service may commission its own recordings, which gives it a consistent house style. It may licence work made elsewhere. Or it may host what creators upload, which produces the widest range and puts the moderation burden on the platform.
The model matters to a listener because it decides how much new material arrives, how uniform the quality is, and whether the person performing is being paid by the service or by you.
Synthetic voices
Generated narration is spreading, and it sounds good enough that a listing will not always tell you. It is cheap, so it tends to arrive alongside large catalogues that grew quickly. Where a service says which it uses, that is a real difference in what you are buying, and a service that says nothing has usually decided not to.
Finding a voice you like
Start with the free end, where performers post work publicly and tag it heavily, and note the names you return to. Then check whether the paid apps you are considering credit their performers at all, because a subscription to a catalogue you cannot navigate by voice is a worse deal than its title count suggests. Sample length matters too: a service that lets you hear a minute before paying is telling you it is confident.
See also audio erotica and the glossary.