On the subject of ISRC codes I've spoken to some of the production people about this Apparently they're supposed to uniquely identify the audio object So if a track appears on a single and also appears on an album with EXACTLY the same recording it should have the same isrc code BUT apparently labels often prefix the isrc with something to identify the label so if a release is re-released on a different label the tracks on it often have different ISRC codes even tho they're the same audio object Basically they're not guranteed Anyway, this is what i've been told... i'd be delighted to be told different... [and they're also no good for describing live music] We have come across gracenotes but unfortunately (once again) they don't really model "the platonic ideal of a song" (just tracks) so mike flowers pops wonderwall http://www.gracenote.com/prof/music/album.html/ambdreampop/7fc4bbab6b367 527a59404978be5b833.html has no "song" to tie it back to oasis's wonderwall http://www.gracenote.com/prof/music/album.html/britpop/6ffbeca624a0d776e 294e04ece5219d9.html they just happen to label the track search as song maybe there's more going on under the skin of the site but i doubt it
________________________________ From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Richard P Edwards Sent: 26 January 2007 01:27 To: [email protected] Subject: Re: [backstage] Music, (meta)data, musicbrainz and the BBC James, The 128 character description could well be the ISRC code from the original label. If it is, then it contains a lot of those same details, and is unique across all manufactured CD's. I would also be surprised if you haven't come across these guys.... http://www.gracenote.com/prof_home.html They seem to have the Song ID database sown up. RichE On 25 Jan 2007, at 16:55, James Cridland wrote: Michael, Ignoring for a while the question of why the BBC is now looking at putting third-party music information services out of business, and being constructive: The major problem we've found working with any third-party music data is the issue of non-standard descriptions. Take a well-known song, which is in our system as... "The Beatles: Norwegian Wood (This bird has flown)", aka "Beatles, The: Norwegian Wood", for example. Life gets harder with R.E.M.'s "End of the world as we know it (and I feel fine)", since R.E.M. is also known as REM and R. E. M. and... ooh, it's horrid. This needs fixing. Secondly, working with third-party systems is a little difficult for cleared-for-broadcast stuff. Oasis's "Fsucking in the bushes" won't look great on scrolling DLS, however we do it - and automated swear filters don't work cleverly enough. (I've added an extra letter in there for work-safe email). The way we've ended up working with these types of services is to have to pre-moderate everything before importing, which is a nuisance but the only way. Easy for us, given the comparatively small amount of music we play; harder for the Beeb, I'd guess. If it helps (which I doubt it will), if you go to http://nowplaying.virginradio.co.uk/vr.js - do it in Firefox so you can see it on-screen - you'll see the following information within a JavaScript line: Artist name ~ artist ID ~ Track name ~ track ID ~ Live on-air studio ~ Presenter name ~ Presenter image reference ~ short description of show (which makes no sense right now I notice!) ~ Short legacy web action description ~ Webcam true/false flag ~ DJ show link ~ Official artist website ~ tickets available true/false ~ 128 character description ~ some number which probably does something I appreciate this is nothing to do with what you're asking, but I wondered whether it was interesting to the conversation. And I'm always up for a pint. j -- http://james.cridland.net/ http://www.virginradio.co.uk/vip/profile/bigjim/
