I assume as well as any VoIP system, which (in my experience) is "meh", although the results can be very tolerable with a little knowledge, care, and post-processing.

An ideal system would automate a "double ender"[1] setup, streaming low-bitrate voice for the conversation itself but also recording uncompressed/losslessly-compressed audio at both ends, then delivering high fidelity sync'd audio as one file per participant (for easier editing) to the podcaster.

[1] https://en.wikipedia.org/wiki/Phone-sync

That is the first thing that I thought about. I know a lot influencers (political and not) on telegram. They can host live radio shows (and record it and call it a podcast). It's great. I'm curious what the community comes up with these new features.