Interesting. What you are saying would indeed be nice but in thinking about this globally ...
I agree, although I represent probably .00001% of users of digital audio, that meta-data is tremendously important, and will become more and more so. I have a medium to large collection of about 7500 albums. I'm OCD about tags,otherwise I even get lost in my own collection. I believe that after the dust settles and the streaming companies that survive and do not go bust, will offer large lossless catalogs depending on very good meta-data that "good" players will be able to retrieve, making (eventually) local content passé. (eg. use JRiver to link direct with Tidal or Qobuz and be able to catalog/tag my own library, using their "leased", cloud-based inventory)
One of the problems is that there are no standards for meta-data - especially for classical music. For popular music, using auto-loaded genres are unusable for me (I either don't agree with them or they are auto-populated by users that don't follow standard values. actually can't follow them even if they wanted to as they don't really exist -- compare discogs to Allmusic) So although Roon may be very good, there are a lot of limitations using it on a daily basis. Auto-tagging is only as good as the data you retrieve; if there is no standard, one ends up re-tagging anyway, unless you only use one source/one application for all your playback and control needs. It seems that some fields have become de facto standards because of iTunes' initial penetration of the market for Pete’s sake
I was disappointed when Tidal refused JRiver's collaboration (or perhaps their terms of their collaboration were too onerous , read $$$??). It would have been “relatively” simple to populate a JRiver library with direct links to albums/tracks/performers .. using JRiver as the control point/renderer and not Tidals interface. But that's the way it goes.
Other than no real standards for genres, sub-genres, and styles - there is not even a standard for album artist, artist, soloists (performers), band members? And the "musicians" as you call them … sometimes you see feat. in the name. does that make it a multiple artist album? It depends .... so doing anything automatically is going to be tricky even with the most basic fields. What constitues a musician for you maybe a soloist for others, an artist for someone else.
But you can compare all you want, just compare apples with oranges. Its also, really about MONEY. I use dbpoweramp, rather than rip from JRIver. Although I like some of the features that it has that JRiver doesn't ... JR rips fine, its included, nothing wrong with it. I bought and keep using dbpoweramp because of their meta-data from Allmusic, Sonata, and GD3. It a once off payment
Roon has access to, it appears to me, much more extensive meta-data ... and you PAY for it. JRiver you pay what ... 50 bucks or less? Add 20 for an upgrade every so often and you get a LOT. Go to Roon and you pay 119 per YEAR (or 500$ for a lifetime). BUT only on
1 machine! You are stuck with their UI etc. That meta-data is going to cost, big time, and I sort of doubt that a kickstarter would garner enough support, it would have to be huge and the fee structure could change over time. Paying extra to subscribe individually to Rovi or whomever controls the meta-data after an agreement with JRiver might be interesting though. (more than interesting
)
Your 3rd suggestion exists in JRiver, but it is not automatic. All you do is create a field like Work, transfer using =[Name] in the field, and clean up the two fields using the replace command. This is a real drag doing 2000 cds, but not for the 10 or so cds most would be adding per month. Again no standards though. Some call a classical work by name including the Opus number; some subdivide the opus number into a distinct field; some use BV others BMV for Bach works – I don't think purely automatic is easy to achieve without getting a bunch of duplicates (not to mention language differences .. should it be the Magic Flute, La Flute Enchanté ou Die Zauberflöte? – probably it should be the latter, but I doubt there would be a consensus for this)
As for delimiters, I agree I'm not a big fan of how JR handles multiple entries and would prefer more of a relational set-up, but not sure this is what you are referring to. Paying a percentage based on hard-ware sales (like per "box") I'd think easier to negotiate than open ended software licensing, and who wants more boxes. And then you are tied to particular meta-data suppliers, who use field names, separators the way they see fit. Again no universal standards.
I would be very interested though in paying for better meta-data -- maybe as a Service plug-in where tags could be accessed form lets say AMG without using Mr. C's Perl scripts, which in practice were just to unwieldy for me. That way it would not be a burden on JRiver financing, the meta-data provider would benefit, the subscription cost could be kept very low. Then this could then be automated and even if the user is OCD like me, JRiver can easily reformat or display this information like you want. I think though this should be after ripping, not while ripping, but there are a lot of couch-potatoes out there
Its really all about consistency
PS - to the OP, if Roon allows you to "auto tag" musicians for example ... what's to prevent you in the meantime to map a custom field in JRiver? Once set-up, after import this would be automatic wouldn't it?