QUOTE (Canar @ 03. Oct 2009, 02:43)

QUOTE (ethex @ 02. Oct 2009, 19:03)

Hm, the .cue-file idea is defenitely not a bad idea. I can take a look at it and see if it's easy to implement.
Shouldn't be too much of an issue. Even if the timing's off, close enough is fine. If it's automated, people don't expect too much anyhow. Even to-the-second would be fine. If you can get sub-second timing, even better. The closer the better, obviously, but I doubt many pieces of streaming software will even give you that data with frame-exact timing anyhow.
I save info about in what exact byte the tag is receieved from now on, so this is pretty much as accurate as you can be with these kind of services.
It's live in the ripper but I have to implement something useful on the web. I dont have much time this weekend, but I'll propably code a little anyway

. I'll see what I can squeeze out!
QUOTE (Canar @ 03. Oct 2009, 02:43)

QUOTE (ethex @ 02. Oct 2009, 19:03)

If you see on some sets the same tag gets repeated, and they even get wrongly sent to other djs sets. I've tried to come up with better solutions to better associate the tags with the right set, but it's really hard to get it fully right :/
IMO the logic here isn't hard. If the track data is the same as the last DJ, ignore it and replace it with the Set Name. When it changes, add a new cue-sheet entry. If it doesn't change for the whole set, ignore the cue sheet.
Yeah, the problem is that if the client gets disconnected and reconnect, the last tag is sent, even though the current dj never set it. But as long as I dont restart the client and it does not disconnect (which never happened yet, but one can never be too sure with streams) it wont be wrong from now on.
But you're right, the rest is pretty simple logic.
If you've got any more ideas like the CUE-file, please share! The programming side of me haven't had this fun for a loooong time