I also share the privacy concerns - it’d be too high a bar of publicly identifying yourself to really encourage conversation (and then I’d be worried about the perhaps non-technical people who would accidentally use it not realizing how their voice is PII).
I’m intrigued by the idea from a UX/convenience perspective though. Maybe you can get the best of both worlds (pseudo anonymity + speak-to-participate UX) if you “salt and hash” the audio consistently? That is, every user ID would have their voice altered in a consistent but irreversable way, that’s also not vulnerable to some future rainbow table~esque attack.
The nosystemd site really doesn’t make a good argument by listing outages… that feels very arbitrary at best (if I showed you our outages at work, you’d think every system was useless; eg: a count of outages could just be a sign that it’s heavily used). I’m surprised the main website itself doesn’t try to at least inline a summary of main arguments against systemd. Since it fails to do that, could someone share a very short synopsis of where the dislike comes from? Or link me to the next summary you’ve seen?