Interesting thread, in general agree with all said, a few things to add
I'm using AudioServer for 3 months now - full setup info below. Mostly use for Spotify, some internet radio, plus spoken announcements/warning paging (slowly building these up).
The quality of the analog speaker driving section is excellent, far better than I would expect for a low power device (as in, genuinely draws very little power) device that's just fed by CAT6 gauge wire.
I do have issues with it though
1/ Poor digital streaming quality. For the large part it's OK but occasionally it glitches - a clicking /popping noise, typically comes in a succession of 3-5 pops. Sometimes goes hours without, sometimes comes in a storm throughout a track. Definitely more noticeable on quiet tracks (e.g. soft classical / evening piano etc). Possibly actually occurs more on quiet tracks - but I think it's just more obvious.
Happens whether with linked groups or with a single zone playing alone. Happens on spdif output and direct speaker output. I need to test built in mp3 from USB/sd card sometime for comparison.
Doesn't happen with Spotify via other sources (e.g. chromecast) to my amp.
2/ Flaky streaming. Once or twice a week spotify just stops playing - error reason "because Spotify stopped providing data". Press resume and it's fine. I feel it could auto retry a bit before crapping out.
It also went through a phase of complaining the account was already in use on another device (when switching who was pushing music to it via Spotify connect) but it seems to have stopped doing that again.
3/ Local SMB network file streaming is a faff. I tried a few times and never got it to connect. I setup my first Samba file server over for 25 years ago so not exactly a novice - just need to find some time to work through it methodically, but it definitely isn't a simple setup
4/ The AudioPlayer is a bit limiting to use in Config - in particular it has no function outputs to give a clue what it's currently playing - I find there's a huge difference between automatically played music, user-initiated music, and paging (at various levels of priority from "put the rubbish bins out" to "the house is on fire") and I'd like to differentiate these states in downstream automation logic e.g. on sound routing and volume, so I had to invent workarounds using other state as a proxy for this information
Now here's the rambling bit about how exactly I'm using it...
My setup is a hybrid of Trey's options. I already had an excellent Anthem MDX16 multi room amplifier, so using this for most of the house, but have setup a Stereo extension for the office speaker (partly because TreeTurbo routing was easier to there, partly as I wanted to get experience with multiroom audio the Loxone way, and partly as that's a sufficiently distinct space that it doesn't alway want to participate in the multiroom audio of the rest of the house.)
For the most part audio server streams digital output to MDX16 and that powers the rooms (inc a Sub for in the Living Rm). If the Chromecast or Nvidia Shield turn on/play they take priority over the Audio server music streaming... HOWEVER if the audio server has a priority announcement to make then this takes precedence over everything - at least that's the idea I'm trying to work towards, it's not reliable yet due to issue #4 above.
The good thing supporting this though is MDX16 has a comprehensive api easy to poke over TCP, to control input selection volume and power all from the miniserver (The Chromecast & nvidia are not so simple, but I have helper scripts in Home Assistant for them which have become very robust in the last year or so)
I'm now leaning towards just use Chromecast Audio (or something like it) for user music streaming, and keep the Audio server just for paging announcements. This is fine, except for the Office that has to rely on Loxone for everything. I could try and route the aux output of MDX16 back into the Audio server input, I need to use a D2A converter (why oh why no digital input, Loxone) and then over treeturbo to the office stereo extension, but so far my brief attempt didn't actually get this to work. I'm also nervous about the high chance of feedback loops this introduces. Also I've become increasingly unhappy with Chromecast stability (high latency and failure rate when trying to connect app to start playing); in comparison the AudioServer is a delight in e.g. triggering a favourite ID from app/ config, it immediately starts playing. So automations triggering music is far better via Audio Server.
One thing to note is the Loxone active speaker range ("Master-client" devices) have Bluetooth inputs, which I believe can be routed over TreeTurbo - so this does open up another way to get more sources into Loxone domain. I would never use Bluetooth by choice though, definitely not for sound for TV (lip sync) or multiroom (sync) use cases. But it maybe useful to others
So anyway, tldr if doing multiroom routing outside of the Loxone ecosystem I'd suggest doing everything outside of it, just use Audio Server for occasional paging purposes, and don't try and mix the two or use TreeTurbo as your PCM audio distribution network.
This is much easier if you already have a streaming source you're happy with, and especially if you only stream local files (I might yet go back to my Squeezebox as the main streaming source, lol. I should probably give a WiiM a go too - been looking at it since before I got the Audioserver, tbh)
Sorry that was a ramble but hope some of it is of some use!