There are a few audio games like this on Amazon’s Alexa platform. When I was looking at it a couple of years ago, there was a level a framework for building them with such narration. I don’t know how you’d work in the multi user angle though.
the framework would need to be designed for that. i'd separate audio handling from actual game logic, so that you can plug in any game logic you like.
you have voice input that gets parsed and interpreted (and also shared with other players, like any voice chat)
and you have audio output, that can be called up like any text messages in a MUD: play this narration. play this sound, etc...
code wise i'd call narration and speech like language translation: say("hello traveller") would call up the audio file that contains "hello traveller". then have the audio team go through all text bits like a translation team does.
