As an update to my problem about grouping music compilation albums together, I've decided to give up on automatic classification techniques and force a more 'robust' solution that will work across multiple SW platforms. I will assign either of the following labels to the "Album Arist" field. I've listed some examples here so others can use them:
For compilation albums/tracks ripped/downloaded from conventional/commericially available media and so on, Album Artist field will start with 00:
001 Multi-Artist Comp.
002 Soundtrack
003 Multi-Artist Tribute
or if you require more fields, you could split them across media type or whatever:
001A Multi-Artist CD
001B Soundtrack CD
001C Tribute CD
002A Multi-Artist DVD
002B Concert DVD
003 Audio Books
and so on...
For compilation albums/tracks downloaded from various sources, Album Artist will start with 01:
010-01 Dld. YouTube Concerts
010-02 Dld. YouTube Fragments
010-03 Dld. YouTube Amateurs
010-04 Dld. YouTube Lost Classics
010-05 Dld. YouTube Car Sounds
010-06 Dld. TouTube Sound FX.
010-10 Dld. Vimeo Concerts
010-11 Dld. Vimeo Car Sounds
012 Dld. Test Tones
013 Dld. Misc. Binaural Rec.
For tracks that I've recorded personally like from FM Radio or direct from a live recording session, they'll start with 02:
020A Rec. FM 95.0
020B Rec. FM 98.0
020C Rec. FM 103.6
021 Sessions 2013 Jan-June
021 Sessions 2013 July-Dec
and so on.
The reasons why I went with the above:
1. I wanted my music to be tagged and classified as robustly as possible across multipe platforms and interfaces. The album artist field should be understood by almost all popular SW and HW out there. I use iTunes, EAC, a Squeezebox Touch with SB Server SW and have two iPods.
2. By having album artist field starting with 0, browsing music should be easier with my iPod and Squeezebox Touch. Using "Category" like "Cat.01 Multi-Artist CD" would have sounded more proper, but then I would have to scroll through a whole buncy of "Catxxx" before I get to artists starting with Catalina and beyond. Similarly, just using the category name without an index number would be worse. I might also have forgot whether I classified them as "Multi-Artist CD" instead of "Various-Artists".
3. It has the additional benefit of separating/classifying and I can browse my compilaton/miscellaneous albums just by entering 0 into the Artist field on my iPod or SqueezeBox Touch, iTues or MC19 and other SW/HW.
4. I have analogue recording hardware and record in DSD from FM radio, occasionally direct from microphone also from DVD/BD-movie discs and so on so the classification benefits of this tagging scheme is obvious.
5. The naming nomenclature also allows adding future categories without upsetting the existing heirarchy. Some foresight and planning is needed though. For example, by using "010-0x" for YouTube downloads, I'm assuming that there will be no more than 9 different categories of music/audio that I'll get from YouTube or I'll intrude into Vimeo's index numbers. In that case, I should use triple digits or switch to letters.
6. My girlfriend is not very tech-savvy and anything that's in her iPod is placed there by myself, so I have to make sure the conventional ways of browsing for music in her iPod is still retained and easy enough for her to use.
Hope the above helps anyone who might be at pains to have their music classified properly across different media platforms and keep in mind that the above method should only be used where there isn't a clear "Album Artist". To me, a clear Album Artist label exists if the name of the artist and or collaborating partner is on the spine of the CD or main heading of the album.
Notes:
If there are people new to ripping CDs and organising/tagging their music, I can only advise them to follow a more source-oriented way of classifying their music with the following assumptions:
1. That you have a large enough music collection from various sources and compilations -Large enough that you will end up having two or more of the same track but from different sources,
2. That you care enough about sound quality to want to know which 'album' or source your track comes from,
3. That you want to keep dulicate tracks from different sources and or
4. You're anal enough (like me) to hate inconsistant music classification/tagging.
I don't think my tagging method applies well to people who only download/rip individual tracks (instead of entire albums or discs), who don't have the time and not concerned about where their music's original source was or who are just not concerned about having two or mroe of the same track or who will just delete all duplicated tracks even if they come from different sources.
To explain further my preferences: I have lots of compilation albums and am bothered by the differences between songs that are supposedly written from the same album but end up sounding different, again due to soruce differences.
Another example: I enjoy the song "Amore e Musica", sang by Russel Watson. I first heard the song after recording it from my FM radio. It's quite a nice and overbuilt vintage radio (Accuphase T-100) and my recorder was a DSD enabled Korg MR1000. Still, the song was quite obviously compressed, but nicely musical and emotionally engaging. I liked it so much that I went and bought the CD. I was horribly disappointed. The CD version, while more transparent and less compressed, was sterile, a bit lean and un-involving. I very much preffered my radio-sourced verison and so that version will be under my album of "Rec. Radio FM92.4" along with other songs I recorded from that station for example.
Now if you've just downloaded the song from some miscellaneous website on the internet and just want to know which album it comes from, then you're not going to be bothered about the source and this naming system is not going to be userful to you. On the other hand, if you're like me and have four versions of "Turn Back the Clock" by Johnny Hates Jazz and plenty more examples like that, then you might want to classify them by the physical CD album in which it came from or the downloaded source or like me the analogue recorded source. One observation I've made is that songs sourced from their originally-released albums (espeically in the early 90s and a bit earlier) tend to sound 'better' to me than ones that come from a multi-artist compilation like "Best hits of the 90s" or something like that. In any case, proper compilation albums will have tracks that are tweaked to equal loudness levels among other processing techniques (some of which can make the songs sharp or lacking body or smoother and more plesant) and so you might want to keep track of which source-album/compilation that song came from.
Yet, you might not want to just keep the version that you like and delete the rest. If you're like me (who came from the days of playing music by disc) and who will listen to music by progressing through the album rather than jumping to songs from all over your music collection, then you may appreciate how songs of the same physical album-CD have reasonably compatible loudness levels. Even if you play music back using volume normalising techniques, consdier too that artists and even the people in charge or arranging the songs in a multi-artist compilation album have arranged their songs with a certain flow or order. Quoting from the book High Fidelity (on how to arrange songs in a mix-tape) "A good compilation tape, like breaking up, is hard to do. You’ve got to kick off with a corker, to hold the attention, and then you’ve got to up it a notch, or cool it a notch…" and so on... I personally feel more complete listening to songs by the album and so will stick to the above naming methods.