Separate different versions of the same song on scores

[arleas]

I bet this has been suggested before, but I don't think I've seen it and this is a simple request that would make things a lot better.

Here's an example from just a few seconds ago to give you some idea of what I'm asking for: I played through the Def Leppard song "Rocket".  I come out with a very respectable end score of over 100k points, getting a clean finish and match 21 only to find that 100k is the MINIMUM score for the top global scores. 

Now I could accept this if it wasn't for one thing:  I played a 4 minute long version of the song, and everyone else is playing the almost 7 minute long version of the song.  Those two versions should not even be on the same scoreboard. 

I understand that differences in the way the songs were encoded can result in a difference of a few seconds here and there on a song. However, when the difference is more than 15 seconds, it REALLY belongs off on its own scoreboard. People playing the standard radio edit version should have a shot at getting on the high score list without worrying that someone with the super maxi extended cut DJ club version that goes for 7 minutes is competing on the same score list.

You don't let Street legal cars race vs Indy Race cars, you don't have heavyweight prize fighters going up against featherweight boxers, and it's not really fair that you have the extended mix people going up against the radio edits just because they didn't mark their MP3 as such.

All I'm asking for is that you provide an option to separate the scores by their times so that everyone is judged vs the same version of the song. 

[Laserrobotics]

It's been suggested before.  Many times.  And the problem is not that people aren't marking the song as such, the problem is that the songs aren't supposed to be marked as such.  They are supposed to have the same name, therefore that's how most people are going to have them tagged.  Scoreboards will indeed tend to be dominated by the more popular version.  You may think this isn't asking much, but apparently things that modify the scoreboard view are either hard to make, or Dylan hasn't gotten to it yet.

[Melodia]

There's no good solution. Any way of handling it would be problematic toward something else.
If you seperate by time, then it'd also seperate ones where a second or two are different simply because they are on different CDs (but still the exact same recording), or whatnot.

[S.]

There's no good solution. Any way of handling it would be problematic toward something else.
If you seperate by time, then it'd also seperate ones where a second or two are different simply because they are on different CDs (but still the exact same recording), or whatnot.

Make it seperate when it has a bigger difference in time.
Another case similarly like this is: some songs are live. Which means that there may be a crowd yell at the end, which sometimes means: BLocks BLOCKS BLOCKS 100k Extra oeh I pwn.
That would also be unfair.

Also, for some reason, I almost beat kEn on a downhill song, only to find out kEn had an uphill version of the song. Eveyrbody else had it too. I Wtf'ed, only to find out that mine was a different version (electric guitar vs acoustic or something like that, I forgot). Now it looks like I ash-hacked, but that's not the case. Forgot wich song it was though.

This gets unfair (dis)advantages, just like your case: Duration. This could be fixed if the difference between tracks is very different. I just dont know how .ash files work, so Im not sure bout that.

Edit: For those interested, the song is Dream On by Aerosmith

[Mincus]

I think there is enough information in the system to differentiate for a large number of these, the question is whether it's worth it?
Starting to put checks like these in leads to a bit of a never ending cycle. There's always going to be an instance where one user or another comes on the forums complaining of this issue. No matter how rigourously things are checked, some will slip through.

So where should the checking stop?
Should it just be on time checks?
Should it include checking ash files?
Should it analyse the song for recognisable patterns like crowds cheering at the beginning/end?

All are plausible, the first is fairly simple, the second is possible but would eat into CPU time at the server end (where bottlenecks occur occasionally already) and the third would require sending several seconds of the track to the server and is clearly over the top.

Just 3 simple examples, but I hope they make my point.
Personally I do think the length should be checked as it would be comparably trivial to the other two (although might require reimplementation of the scoreboard to cater for the length differences).

[lavacano201014]

...reimplementation of the scoreboard...

I have a feeling this is the only way we can stop these "OMG SCOREBOREDS R UNFARE" threads.

[arleas]

If the scoreboard already separates people who use tags such as [as mono] or whatever (i am not familiar with all the tags or how exactly they are used) it should be trivial to separate them by the song length.

Most of the songs that are only different by a few seconds are only different by a few seconds because of added silence (which doesn't add any blocks but gives you time to rearrange your blocks or maybe get a new random powerup on modes like vegas).

 I would rather see a scoreboard that accepts 10 seconds variance in either direction from the original than one that allows 7 minute and 3 minute versions to be put on the same scoreboard. Actually I would rather compete vs all the people who used the same exact version of the song I'm on than to compete vs people who have merely named their track the same as mine (which is about what this amounts to) but without some way of fingerprinting tracks that would be too difficult.

In the extreme case, you could use a fingerprinting scheme similar to musicbrainz to determine if two songs were identical or not... hell, if you could even match the song length + the track outline to each other that'd be better than what we have now (which is nothing).

And again, putting people on their own scoreboards shouldn't be THAT extreme of a solution as to call this impossible without totally revamping the scores.  It's just extremely annoying to see that the only way I could come close to the high scores on the list would be to go into audacity, copy the entire song, paste it to the end of the song so it's now double length, and THEN go play it. 

[Mincus]

The tags just change the title.
To the Audiosurf score data base it just sees it as a differently titled track.
This can be easily demonstrated by just swapping tags around: [as-monoonly] [as-nostlth] gives a different scoreboard to [as-nostlth] [as-monoonly]
Adding in time differentiation means adding in extra checks to the score database as the scores come in, but in addition to that, every score that is currently up there would need to be checked and shuffled to the correct database.
It may be that the database was never designed with enough flexibility to allow an additional variable to create new scoreboards.
Also, how do you sort existing scores?
Do you just take the shortest version of the song in the database and go up every 10 seconds?
Do you take the most common song, go 10 seconds either side of that and split out other versions from there?
Do you start with the longest?
It's a lot more complicated than "just create new scoreboards for different length songs".

I'm not saying it isn't possible, and indeed the checking itself is trivial, but depending on how the database is arranged internally it could be a lot of work.

[Melodia]

The worst thing is that splitting them in the various ways suggestion won't help the opposite problem -- in cases where the scoreboards are split because of different names. Even a "the" will make different names, but especially with video game music the same track can have a wide variety of both artist and name entries:

For instance, all of the following are the same track:
Nobuo Uematsu - Terra
Final Fantasy III - Terra
SQUARE - Terra
Nobuo Uematsu - Tina
Final Fantasy VI - Tina
SQUARESOFT - Tina

I can imagine people using any of those combinations, as quite often people will use the game or the game company as the artist. Granted with FF most people know better, but it was a good example of a track that has two seperate but official names, as well as two seperate but official game names....so yeah.
I'm sure I've missed out on compteting with many tracks because of this.

[Sensei Le Roof]

The worst thing is that splitting them in the various ways suggestion won't help the opposite problem -- in cases where the scoreboards are split because of different names. Even a "the" will make different names, but especially with video game music the same track can have a wide variety of both artist and name entries:

For instance, all of the following are the same track:
Nobuo Uematsu - Terra
Final Fantasy III - Terra
SQUARE - Terra
Nobuo Uematsu - Tina
Final Fantasy VI - Tina
SQUARESOFT - Tina

I can imagine people using any of those combinations, as quite often people will use the game or the game company as the artist. Granted with FF most people know better, but it was a good example of a track that has two seperate but official names, as well as two seperate but official game names....so yeah.
I'm sure I've missed out on compteting with many tracks because of this.

Oh, it gets worse. The way I have the song tagged, it would be on a scoreboard for "Final Fantasy series - Terra". (Though I have them tagged such for my own Zune-listening convenience.)

As far as previous sub-topics, I also think "extended mixes" should be split into a different scoreboard, since enough data is added to make a significant change. You wouldn't challenge 9-minute "Free Bird" scores with the 4-minute version, would you? (Back off, it's just an example.)

[lavacano201014]

I have the ~3:30 version of Rammstein's Amerika, and there's a 6 minute version (that I want - I like long versions better, usually). But the top of the scoreboard is the 3:30. So, it's possible, just very damn hard, methinks.

[arleas]

The tags just change the title.
...
Also, how do you sort existing scores?
Do you just take the shortest version of the song in the database and go up every 10 seconds?
Do you take the most common song, go 10 seconds either side of that and split out other versions from there?
Do you start with the longest?
It's a lot more complicated than "just create new scoreboards for different length songs".

I'm not saying it isn't possible, and indeed the checking itself is trivial, but depending on how the database is arranged internally it could be a lot of work.

Depending on how the Audiosurf tracks are created, those could also be "fingerprints" to determine if the songs are the same.  That's the other thing that annoys me (when people have a completely different Audiosurf track from everyone else and they're on the scoreboard).

The thing is, if everyone used musicbrainz, then all these extended mixes or live versions would be marked as such in the title, or at the very least, in the album.  Does Audiosurf use the album information? 

It could potentially be separated into different album versions (provided that people actually have their crap tagged properly). 

It could still be an option to group them all together by name so that people who don't care would see the scores exactly as they are now, but I'd rather group them by who is playing the exact same song rather than by who is using the same name on their mp3 (which may or may not even be the same song).

[Riffraff]

I bet this has been suggested before, but I don't think I've seen it and this is a simple request that would make things a lot better.

Here's an example from just a few seconds ago to give you some idea of what I'm asking for: I played through the Def Leppard song "Rocket".  I come out with a very respectable end score of over 100k points, getting a clean finish and match 21 only to find that 100k is the MINIMUM score for the top global scores. 

Now I could accept this if it wasn't for one thing:  I played a 4 minute long version of the song, and everyone else is playing the almost 7 minute long version of the song.  Those two versions should not even be on the same scoreboard. 

I understand that differences in the way the songs were encoded can result in a difference of a few seconds here and there on a song. However, when the difference is more than 15 seconds, it REALLY belongs off on its own scoreboard. People playing the standard radio edit version should have a shot at getting on the high score list without worrying that someone with the super maxi extended cut DJ club version that goes for 7 minutes is competing on the same score list.

You don't let Street legal cars race vs Indy Race cars, you don't have heavyweight prize fighters going up against featherweight boxers, and it's not really fair that you have the extended mix people going up against the radio edits just because they didn't mark their MP3 as such.

All I'm asking for is that you provide an option to separate the scores by their times so that everyone is judged vs the same version of the song. 

Anyway, you should be the one labeling it (radio mix) or (radio edit); Very rarely are radio versions the "first" or "normal" versions of songs.

[Mincus]

Depending on how the Audiosurf tracks are created, those could also be "fingerprints" to determine if the songs are the same.  That's the other thing that annoys me (when people have a completely different Audiosurf track from everyone else and they're on the scoreboard).

It's too "precise" it would need to be fuzzied in some manner for this to work properly, otherwise we'd get a couple of scoreboards for the pirates, one for the FLAC users and almost everyone else would be on their own depending on what program/settings/format they used to encode the track. I don't doubt the fuzzying is possible, but it would be a lot of work.

The thing is, if everyone used musicbrainz, then all these extended mixes or live versions would be marked as such in the title, or at the very least, in the album.  Does Audiosurf use the album information? 

MusicBrainz fails horribly on quite a lot of my tracks. In many cases it simply doesn't know, in the extreme case it places it in a completely different genre with a completely different artist.
It's possible it could be used for "hinting", that is, if MusicBrainz looked the track up and reported a different tag, then the user could be asked. Or a button could be added on the loading screen for optionally looking the track up maybe.

It could potentially be separated into different album versions (provided that people actually have their crap tagged properly).

Often the problem is different tagging styles. Not everyone wants their tracks tagged the same way.
Take the Final Fantasy example earlier in the post. Some people won't care, but equally they'll be some with each of those tags that will insist that's the best way of tagging them.

It could still be an option to group them all together by name so that people who don't care would see the scores exactly as they are now, but I'd rather group them by who is playing the exact same song rather than by who is using the same name on their mp3 (which may or may not even be the same song).

I suspect this would require significant changes to the scoreboard system, which is something Dylan has been extremely reluctant to do.
Also it makes it very difficult to work out dethroning. Do you send an e-mail for the "don't care" scoreboard or the "precise" one you like?

[Melodia]

The thing is, if everyone used musicbrainz, then all these extended mixes or live versions would be marked as such in the title, or at the very least, in the album.  Does Audiosurf use the album information? 

It doesn't. It's been suggested many times before, but the problem with using album is that many times different albums will have the exact same track. Some tracks may be on an album, a rerelease of an album with bonus tracks, three different compilations for the band/artist, and four "best of genre" types....not to mention more academic history sorts of compilations, and so on and so forth.