Laura writes:
| >>>>> "Frank" == Frank Nordberg <[EMAIL PROTECTED]> writes:
| Frank> Wil Macaulay wrote:
| >> Lets have a little reality check here, and ask ourselves a
| >> couple of simple questions: Why are there almost twenty
| >> thousand tunes freely available in abc format on the internet,
| >> more than two orders of magnitude more than any other format?
|
| Frank> You're underestimating both ABC and "other formats" here.
| Frank> Musica Viva alone has 2192 tunes in GIF format - 313 of
| Frank> them are also in PDF format and 1239 of them in ABC format.
|
| I think Wil was talking about formats in which music *information* is
| available, not formats which graphically display information. All the
| stuff in GIF or PDF on Musica Viva comes from music information which
| has been entered in some other format, such as ABC or finale.
This is correct, and quite relevant. GIF and JPEG aren't musical
formats, and no musical information can be extracted from them. There
are a lot of images of printed music on the web in such formats, but
there's nothing you can do with them except download them and print
them. Some of them came from scanners, and some via conversion from
true musical files. In a couple cases, people have scanned in music
so that others can transcribe it to abc. (This isn't unusual; it has
been done with a lot of old documents, most notoriously with the Dead
Sea Scrolls.)
I've tried to find a way to locate images of music on the web, and
it's pretty much hopeless. I've succeeded with abc because it's easy
to extract musical information from abc files. I looked at the
songwright and nottingham formats, which each have a few thousand
tunes on the web, but most of them have been translated to abc, so I
gave up on that. (A bit of a pity, I'd say, since it wouldn't hurt
abc to have a bit of competition. But I understand why they did it.)
I wonder how feasible it would be to add other formats to my abc
search bot? I've glanced at lilypond and MusicXML, but not long
enough to grok how I might write code to extract information such as
titles, keys, etc. I looked into adding midi to my index, but a quick
check showed that very few of the .midi files out there have a title,
and that's what I use as a key.
(Curiously, there are a couple of abc sites whose files lack titles.
One is http://shiva.di.uminho.pt/~jj/musica/abc, which has a bunch of
interesting-looking trad Portuguese tunes. My code doesn't index
them, either.)
BTW, the last run of my search program, a few days ago, gave the
bottom-line numbers:
Tunes Titles Files
80209 89310 25135
The are counts of the X: and T: lines, and the distinct URLs. I've
been lax about posting such numbers lately, and I should add the
usual disclaimer that they mean less than you might think. There are
around 25,000 distinct titles, with a mean of betweeen 3 and 4
occurrences of each. But even this is suspect, because of spelling
variations. Some of the replication is from different versions of
tunes, but most is because of mirroring.
Still, I'd estimate that there are maybe 20K truly distinct ABC tunes
on the web that my search program has found, on about 125 machines.
I've found a couple dozen new sites in the past month or so. Most of
them are small, but there was one site with over 6000 tunes (mostly
copied from other sites, so this site qualifies as a "mirror" site,
and contributed very few new titles.) I'd bet that there are a lot of
other small ABC sites out there, and there's a small chance of
another big one. If you know of any obscure small abc sites, you
might go to my finder:
http://trillian.mit.edu/~jc/music/abc/findtune.html
Type in the host name, and it'll give you a list of all the known
tunes on that host. If it's not found, send me the URL.
Is this feasible with any other musical formats?
To subscribe/unsubscribe, point your browser to: http://www.tullochgorm.com/lists.html