Earlier quoted context omitted.
Not GP, but Discord user too. UX matters. People are very sensitive to the context of features, what the subtle meaning of using them is, how we initiate the communication and how it all "feels" like, for lack of a better word. Kind of like there is no difference between (s)ftp and dropbox, and yet there is.
Ok but just for me as an observer, it's not clear what the actual UX difference here is. The discussion seems to suggest that the actual capability of the UX is in fact worse in Discord, and maybe not worth copying exactly if it causes a feature regression. It would help to specify exactly what you would like changed around, i.e. if there is a button that could change to make it more clear what is happening when you…
It's also broadcasting who -is- talking; I've very frequently made a binary choice of "do intrude" versus "don't intrude" based on who's actually talking. I know based on the social circles that if person A and B are talking, they're likely "talking shop", because person A commissions work from person B. But I also know that if it's person A and C, they're probably just gaming, and I'm welcome to join them.
This implicit social signaling is obscenely important. It's a make or break feature.
When I'm using slack, by comparison; I can see that a coworker is in "some mysterious voice call" but I have absolutely NO IDEA who they're calling, and whether I'm welcome to join. I have to openly intrude on their time, and ask that explicitly, and that's inherently rude. I'm able to do that with close friends because I have a huge buffer of goodwill, but interrupting people is always inherently rude.
Discord gives you a way to get this information without being rude.
If the only way to do it is to be rude, then - here's the kicker: It introduces an outright failure state! Quite a few people will actually decided "gosh, I don't feel comfortable interrupting this person I barely know, I'm just not going to." They literally DON'T make a call they otherwise would make, and it's 100% down to UX. It's a full on, binary, "failure to provide service", and it's because of how your software psychologically runs on the people using it.
There are two computers running your software, not one. One of them is electronic, but the other one is biological. UX is about that second wetware system; some of it is calculable, consistent stuff that holds true for a statistical majority of people - just like how certain optimizations will benefit a "consistent majority" of the electronic computers running your software.