My feelings were always mixed with PulseAudio, I loved the general flexibility it provided and the host of tooling for it like PavuControl. It was also the first time I saw a sound server expose the whole source/filter/sink type architecture and I really digged that. But it had these weird issues that would rarely occur, but were a nightmare to fix.

Like I recall I couldn't listen to Doom Eternal's audio over bluetooth due to some latency issue. Tweaking the PulseAudio config would resolve the issue but break other things. Other times it would crash applications, but you wouldn't know it was PulseAudio that was responsible until you dived really deep into it.

It was had this inherit instability to it without any of the liability, since you rarely would identify it as the cause. Since audio is used by most graphical applications, it could be said that PulseAudio was indirectly responsible for increasing the total instability of the entire desktop environment in a very subtle way. Took me years to accept this since I was pretty biased towards it tbh.

Overtime I realized what I liked about PulseAudio was it's flexible API, but the server implementation itself was mediocre. So when I first learned about PipeWire, I was delighted they had ensured it was backwards compatible with PulseAudo, so much so I used it even before it was available in most package managers.

There was this irony that it felt more stable to me than PulseAudo, despite literally being marked as "unstable" by the devs at the time. I took it as a sign the devs were highly committed to quality.