According to the annotations, these are different recordings:
I have been able to verify that there is a “softer” orchestral version (here), but it’s distinctly shorter, about 3:20 (and the only place I can find it on MusicBrainz is a Chris Cornell bootleg). I’ve spent a lot of time on youtube, spotify, itunes, etc, and I haven’t been able to distinguish any of the 4:00
“guitar” versions from each other. Single, soundtrack, solo album etc all sound the same to me. Unless someone else can verify it, I’m going to merge these.
Added: According to wikipedia there ARE three versions. I’m still trying to track down the “Main Version” vs “Pop Mix” for side-by-side comparison, though.
Both single and album version report the same ISRC US-UM7-06-13448, but they are not the same.
Differences are especially noticeable in the beginning of the song. The single version sounds rockier, using e-guitars and drums. The album version sounds more orchestral, using acoustic sounding guitars and brass.
Yeah, that’s what I’m trying to hear. Maybe I just have bad ears. If someone can point me to the two different versions, maybe I can add them to the annotations.
I haven’t listened yet, but will as soon as i can. Thanks!
Update: yeah, now I can hear the strings and horns on the album version vs the single. Thanks. Do you know which one is considered the main version vs the pop mix?
Ok, now that I’ve got their “fingerprints”, I realize that part of my confusion was that most of the singles were pointed to the wrong recordings, because the “pop mix” IS the album version. So I’ve fixed them:
@psychoadept lovely bit of work there. It gets funny how often making one little change to some data on MB can lead to chases down rabbit holes. I’ve learnt a lot about some obscure artists due to untangling confused clashes like that.
Thanks! I’m pretty sure the acoustids are equally a mess (and moving recording around probably doesn’t help), but not sure how to go about sorting them out. Anybody have ideas?
Quite interesting situation here. I’m wondering whether it’s an algorithm of AcoustID problem or it’s just a human mistake? I’ve found somewhere in the internet this release https://beta.musicbrainz.org/release/0c965fd7-18c0-4286-b509-b6da0a44bb2d with both variants and checked the acoustic id.
You Know My Name (single version) has acoustID: 8ee8ce88-6552-45f2-ac2b-60abf3a1df85
You Know My Name (album version / pop mix) has acoustID: cb270744-ef63-4c24-996b-cf55dc3a4b43
Read between the lines here of how some users make use of Picard. They will throw thousands of tracks at Picard that have come from all kinds of messed up sources. Tracks like this will have ended up in collections with just basic tags and names.
Many of those users just don’t care about which version they have. They just want a name and a date. And often the date they want is the original release date and not the actual date of the release of the compilation their track is on.
So a match occurs based on the scraps of tags they have… and then they hit the Submit AcoustID button. ARGH! They have now attached the wrong version of an AcoustID to the wrong track.
This can be spotted when there are a majority of AcoustIDs attached to track “A” and a few attached to track “B”.
This kind of rot will be especially bad on a track like this which must be on hundreds of different compilations now…
What IvanDobsky says. Also, the fact that several releases in musicbrainz were pointed to the wrong recordings will have made it worse, because even if someone was paying attention and submitted exactly the correct release, now those acoustids are attached the wrong recordings. And now that the recordings are assigned correctly, the acoustids will be in the wrong place on the releases.
I think the best we can do is make sure the specific acoustids you have, and I have, are attached to the correct recordings. I don’t think the rest will ever be completely sorted out; it would take someone who knows what they’re looking for getting their hands on all these compilations just to be sure WHICH version is on the compilation (maybe some day I’ll try to track down samples on itunes/amazon/etc to see if I can figure some out - no time right now). Currently they’re pretty much all attached to the album version, which is probably reasonably accurate at least for the Chris Cornell collections.
(For the record, the single version I have gives the same acoustid as yours. The album version is coming up different, though.)
And then add in all the manic merges that seem to happen. Some members of MB merge based on lengths and AcoustIDs which in theory is fine, but in practice means errors like this get built into even more complex errors.
Also add in that when someone is adding in a Compilation disk like “Bond Film Theme Tunes” it can be very boring, complex and time consuming to link all the tracks up correctly to the correct recordings. Especially for a noobie. The GUI is too small to be able to really understand the differences between version of a track in that window, so “close enough” can often occur. Leaving a compilation pointing to the wrong version, and then attaching its AcoustID to that wrong version. I know I will have been guilty of that in my early months here.
IMHO there is no clean way out of this without, as @psychoadept suggests, someone having to physically work through disks. Maybe we need a “certified” AcoustID system? Some kind of update where editors with OCD are allowed to “certify” that their data is coming from a CD and only a CD/vinyl/Original?
Something to add some quality weight to certain AcoustIDs?
I don’t look at the AcoustID site much - but do notice that when a name of the track is included then a counter can lean towards which version it really is. BUT that then drags us back to the same old issue of some people just not attaching the correct tracks…
Certainly the worst tracks are ones like this. A one hit wonder of a track that spreads all over the place to maximise cash-in. And then it is piled into film compilation disks. And Bond disks. And many places that really won’t care which version they have as long as the name is correct. Try reading the credits on those CD booklets and it is impossible to know which version you have in hand.
Why won’t just make 3 recordings in MB: “single version”, “album version / pop mix” and “unknown version / unspecified version / i don’t care version”?
But in this case I think it would be better to rename the current “album version” into “unspecified version” and create “album version” from the scratch without any irrelevant AcoustID.
Don’t know what others make of this, but if it’s messy enough to not be useful anymore I think it’s okay to do a spring clean/remove ID’s, so that people re-submit them (now that the versions are separated out there’s a chance they’re adding their ID to the right one).
Sometimes the only way to move forwards is to take a step back.
Yeah, even the single version had more of the album version acoustids attached, lol. I’ve sorted out the ones we’ve identified, may go with your suggestion for the rest.
One problem with deleting lots of AcoustIDs is you are assuming that those who submit them will know where to attach them. Look at how many compilations this track is on. And notice that it will never say which version of the recording is being used on those cover notes.
You are probably better off moving AcoustIDs around to an assumed better fit than relying on all those compilation albums to be allocated correctly.
The other solution I can suggest is to create new recordings for every compilation where exact version is unknown and leave the problem for the future generations. After some period of time they will attach correct AcoustID to the blank recording and merge them with recording of specific version.
