Thanks for the recommendation.
1 -- composer! It needs a field to itself. That's the most important single fact about any classical track, and yet most streaming platforms ignore it or fail to display it. If you can't unambiguously identify the composer of a track in the metadata, get out of the business.
2 -- entire work, and movement numbers! Most classical pieces have multiple parts meant to be played in a specified order. Imagine playing "Dark Side of the Moon" on shuffle, and you'll know what classical streaming is like.
3 -- performer! Not the composer. Amazon fails this simple test frequently -- they'll list both the pianist and the composer as "artists," as if Glenn Gould shared his piano bench with Anton Webern during the recording sessions. If it's an orchestra, give me both the orchestra and the conductor.
4 -- soloist names! Who's that playing the piano? Not the conductor! Unless it is, and he's conducting from the piano. Which happens.
5 -- year of recording! Sometimes the exact same artists return to the exact same work decades later.
The basic problem is that streaming services default to "track" / "artist" / "album" -- where "track" implies independence from other tracks and therefore the shuffle problem I mentioned before, "artist" could mean just about anything, and "album" as we know it is a concept that postdates most classical compositions and isn't very meaningful, especially when compared to the (unsupported) name of the composer.
Having an entirely separate system for classical is the only way to make it work.
All of this, of course.
The principle trouble with Classical is
inconsistent metadata from the record labels. This goes back to the CD era - pulling track info metadata from CDs and porting it to digital download is a classic example of garbage in garbage out. When I copied my CDs into iTunes (metadata provided by the data encoded on the CD) it was clear that the CD producers often dropped the ball, encoding inconsistent metadata into the digital master from one track to the next. Even a simple misspelling on one track could break the integrity of a four-movement work.
Of course, I can't blame CDs entirely for the mess. After all, many CD releases were re-releases of black vinyl, 78s, and even some Edison cylinders, and back before Edison, when sheet music sales were everything... still plenty of inconsistency on every step of the journey.
Should it be Beethoven's Symphony No. 5 in C minor, Op. 67? Symphony #5? Beethoven's Op. 67 symphony in C mol, "The Fifth?" Beethoven's Symphony in C minor? Etc., etc.
There are further issues, of course. Record labels in various regions of the world using different terminology/naming conventions; the inconsistent use of popular/colloquial names for works; the failure to include composer catalog numbering (Köchel, BWV, etc.).... There are many database fields that can be filled. Without proper input validation and standardization, it's a total mess, and proper database normalization (and search indexing) is a huge task.
Then there's the third-party service Apple uses for metadata/album art. Plenty of room for further compounding of the errors/inconsistencies of the record industry.
That's a substantial part of the job undertaken by Primephonic - wrangling a huge catalog of recorded music into a database that serves-up accurate query results while respecting the specific needs of classical listeners (whole-composition playback, etc.). Clearly, from Apple's standpoint buying Primephonic was more efficient than fixing what it already has.
I hope that Apple does not operate separate Music apps by genre. No genre should be excluded from the larger whole that is music. I have eclectic tastes, I want to access them all in one place.