Frames for track lengths

[Hellacia]

First of all I don't know how ridiculous this is or how much rewriting it would take, this is just something I think would help kill annoying, mass circular editing, and improve the look of things a tiny bit. When editing a tracklist, in the track time field, frames should be included. So, minute:second.frame (example, 7:24.15). Not sure if I need to explain frames, but there would be two reasons to include them.

1: It would help to separate track times that were obtained from a CD source with a reliable program from track times that were obtained from MP3s or the use of programs that round track times up or truncate the frames completely.
2: It would make disc lengths display correctly. Currently all disc lengths exclude frames, so they're usually some amount of seconds off. Not a big deal but it really isn't displaying the correct length and I do think we are about accuracy.

However, in this case, I don't think we should discourage information just because it isn't that accurate. If only general times for tracks are known, perhaps the frames field could accept ?? and then it would just be ignored. We couldn't really accept 00 as "unknown" because 00 is a valid value. Also, I'm not saying this should display in the tracklists on the main entry page, if you wanted you could just make the mm:ss display (or I don't know actually, could you do that? ). I think that would be better when looking at a track time.

I've seen a couple discussions about track times lately and I think this would help end confusion about whether or not certain track times are correct.

[Dag]

This makes sense to me. Proper accuracy is good. (though honestly I know little about frames )
But DB changes are delicate, plus this doesn't help much to regular users, so probably won't be high in the 'todo' list. (open source...?)

[LiquidAcid]

I want to clarify, that the exact number of samples _can_ be extracted from an MP3, provided that it has a proper Xing/VBR/LAME header. So e.g. every MP3 that gets encoded with a recent version of LAME contains the exact sample count. And foobar2000 can display this information and probably a ton of other tools.

One more reason to use LAME over the crappy iTunes codec

[Hellacia]

Yeah that's true, but I really don't think many people that are taking times from MP3s are putting this knowledge to use or even know this. The method of using LAME headers to extract sample information about MP3s is fairly obscure, I think, and the casual listener that is adding track times from an MP3 rip he found probably isn't making use of this. You are right though, it can be done. But then, this is assuming that the source file and resulting MP3 were both taken from a source that was the exact right track length. A number of things can happen when ripping an CD and the resulting file is not always perfect. It also depends on the methods people use to rip CDs, such as gap handling; depending on how people treated the pregaps, the source WAV for the MP3 will have a different time than what is actually the time on the CD. This is another reason why I really wouldn't recommend using MP3s as a source for frame-accurate times even if you do know how to get the original sample count (which I do, just to say). A lot of people don't even know where the MP3s they're using came from or how they were created.

[LiquidAcid]

Yeah, but you've got the same problem with taking times from CDDAs. Any idea how many bootleg rips are floating around on the net, where the owner isn't even aware that he doesn't own a original?

[Hellacia]

You're talking about any random rip you can see floating around the internet? Okay. But I really don't want to go too much into the best ways to download music because I don't think it's an appropriate topic for here.

By the way, just to clarify my own point about pregap settings affecting track lengths: the CD would have to have been ripped to prepend the gaps. Nobody prepends gaps. Every rip you will ever see floating around the net appends gaps (or maybe removes them, but not prepends them). Nobody wants to hear silence at the beginning of the track, but this is the track's native state on a CD. So MP3s are really never to be used to determine frame-accurate times even if you do know how to extract the information from a LAME header. Yeah sure you can extract the exact number of samples for the source WAV, but the source WAV was incorrect, so your time is still incorrect.

[LiquidAcid]

Originally Posted by Hellacia

You're talking about any random rip you can see floating around the internet? Okay. But I really don't want to go too much into the best ways to download music because I don't think it's an appropriate topic for here.

No, I'm talking about what this "floating around" implies. Namely that there are tons of bootlegs in circulation.

I just see that we're opening a can of worms here: If we want the additional accuracy, then we also have to ensure that it really _is_ accurate. How to do this? Cuesheets from AccuRip-verified rips e.g.?

[Hellacia]

Sure, that's one way, though I'll mention that most of the bootlegs you've brought up will check out with Accuraterip because they have the same data with just a different offset.

Another way is to take a screenshot of a program while the CD is in the drive. We could also ask for a picture of the CD used. But then we could just assume those are fake too. If anyone has any good ideas, throw them out there. But, I still think it's better to include the option to add frames. Perhaps trusted editors would only have access to editing them.