Meta’s Threads does this. You can post a voice recording and it transcribes it so you can also read the post. This seems like it would be be best of both worlds for those with visual accessibility issues, and sometimes it is nice to hear what people sound like and the actual inflection of the words verbally.
The converse is most people can already use speech-to-text. My phone has a microphone button as part of the on-screen keyboard and it will dictate what I tell it into text. The only thing is it doesn’t involve keeping the audio recording. I could post this via speech to text right now if I wanted.
I’d casually PM with the user to see if they might use speech-to-text, explaining that text bandwidth is much more efficient than file uploads. It’s possible they’re doing that “memo to me” dictation in the car. If what they’re typing is too long to type, it sounds like they should actually be starting a podcast and linking people to it.
Maybe some forums might warrant a specific category for “audio discussion” if it’s a big thing? Many Discord chats I’ve been a part of have audio chat channels, but it’s rare for people to be in them unless it’s a planned meetup or event.
We have this similar conundrum with Discourse chat sometimes. Users are like WE WANT CHAT and we make a channel, then they’re like WHERE IS EVERYBODY? and the chat winds up functioning basically as weird stream of consciousness forum topic people visit occasionally. (We do appreciate chat and we have our hardcore friend-groups who use it more like a group SMS on their phone.) The OP’s user sounds like they’re doing this, only it’d be like communicating via voicemail which does not sound fun.
As previously suggested, this might self-correct as most users probably feel the same way and if your voice-responder doesn’t get responses, they’ll naturally stop doing it.