theobald nospam at paraskevas.capsl.udel.edu wrote:
> [snip]
>> Since most classical music is public domain, the information about the
>> source music should be separate from the CD track information, since
>> there can be a many-to-one mapping from the latter to the former (more
>> than 1 CD can have the same piece of music). Entries for the pieces
>> should be in a separate table (sorry I don't know the proper DB terms
>> for these concepts). It would be very convenient to have a common entry
>> for, say, Beethoven's 9th Symphony, and the entry for each available
>> recording would link to this entry. This would be more convenient for
>> people entering discs (except for the first person to enter a given work)
>> and cut down on discrepencies, typos, etc.
>> Also nice would be separate keys for conductors, orchestras, and
>> record companies.
>> There would be a glitch with CDs that split movements into multiple
>> tracks or combine multiple movements into one track, but I think this
>> could be handled.
>
> The problem as you present it isn't well-defined. Is Beethoven's 9th
> as played by the Boston Pops the same as that played by the Nigerian
> Wind Symphony?
Depends on the definition of "same." Obviously, two performances are
distinct. But there is a pretty much standard version of the 9th (or just
about any other work) with four movements and a standard notation for each.
95% of the versions sold on the market use this standard version, and all
will label the movements the same exact way, except for slight punctuation
differences (using dashes instead of semicolons, for instance). In addition
to the movement names, all will use the same piece title, opus number, key
notation, etc. This is common information which could be inherited by any
disc containing a performance of the 9th.
There are a few problems with this. One, as you suggested, is alternate
versions. Some pieces have "transcriptions" ("covers," if you will) by
other composers. Liszt, for instance, did a transcription of Beethoven's
9th for piano. But this is already recognized as a separate piece by
standard music catalogs. A tricker problem may be alternate or modernized
versions -- I doubt the Nigerian Wind Symphony is going to play the
standard version unless they augment their orchestra with some string
players! But these can be treated as special cases.
Another problem is track-to-movement mapping. The last movement is
split into a purely orchestral section and a choral/ orchestral
section. Many CDs have separate tracks for them. Some even have
further splitting, since track boundaries do not force pauses.
A final, albeit minor problem, is language. Key names, for instance,
vary from language to language. A 'B' in English is not the same as
a 'B' in German! Probably keys could be represented in a
language-independent form, or choose one language arbitrarily, and
the user's reader translates. Also, Beethoven's 9th is subtitled
("Choral" in English), but again, different people want to use different
languages. (I'd suggest using whatever language is considered "official"
for each piece, which is not necessarily the composer's mother tongue --
Haydn used English names for his London symphonies and French names for
his Paris symphonies.)
> I would be happy to see us make a good stab at supporting classical
> music in a more reasonable fashion than CDDB, but I think that
> starting off with a object in an object database for every song is (1)
> overkill and (2) ambiguous.
Maybe true. At least then the database should allow some sort of hierarchy,
the way Workman does, and there should be clear guidelines.
Or there could be some form of inheritence, like I want to put in my new
recording of Beethoven's 9th, and I find another recording already in there,
so I just say "my tracks, piece name, composer name, opus number and key
are the same as that disc over there; only the performers are different."
-- ------------------------------------------------------------------------------ | Kevin B. Theobald, Ph.D. | | CAPSL -- Computer Architecture and Parallel Systems Laboratory | | Department of Electrical and Computer Engineering, University of Delaware | ------------------------------------------------------------------------------ | Win $200 from Microsoft -- just mention their products in your papers! | | http://chronicle.com/data/articles.dir/art-44.dir/issue-33.dir/33a03001.htm| ------------------------------------------------------------------------------