I’ve been trying to identify and separate different versions of tracks 3-6 from An Evening Wasted With Tom Lehrer, which based on the releases in MusicBrainz I had come to the conclusion were changed with the RS-6199 release in 1966.
I still think that, sort of, but then I thought about a live album on vinyl and how little track length would matter in that context. When I started looking around on Discogs, I found that actually all the Reprise releases appear to use the lengths of the “second” version (contrary to what’s in MusicBrainz). Nothing prior to that even states lengths on the covers, although maybe there was some indication on the vinyl itself of what could be considered a track break. But practically speaking, there were no track lengths until Reprise labeled them.
What’s for sure is that the CD release in 1990 used different track lengths from those stated on the vinyl releases, while The Remains of Tom Lehrer and some other compilations more closely match the vinyl lengths. (These two versions can be heard side-by-side if you download the standalone album and The Remains of Tom Lehrer from tomlehrersongs.com.) I suspect that some of the earlier releases in MusicBrainz borrowed the lengths from the CD, creating the impression that RS-6199 changed them.
I’m considering doing the following:
Update all pre-Reprise releases to have no track lengths, unless some documentation of them can be found
Update all Reprise vinyl releases based on the covers in Discogs (this part makes sense regardless)
Regroup all vinyl tracks into one recording each, along with the later releases of similar length. This would likely leave recordings from the 1990 CD and matching digital release on their own, aside from some scattered compilations (just tracks 3-6, the rest have been consistent).
Good idea? Terrible idea? Other suggestions?
Tagging @chabreyflint since they’ve done quite a bit of work on Lehrer’s releases, too, and have been following my slew of edits lately.
Difficult to prove lengths without a copy in hand. I’ve seen similar problems before. It is common for a release to have the CD track times due to the way the GUI suggests copying another release’s data when adding a new one. This is a problem both here and is worse at Discogs. At least here at MB we have DiscIDs that often fix track times.
Cover art isn’t ideal either. Never trusted those times. I’ve seen reissues with longer track times on a recording, but original artwork showing different times. Had an album like that the other day. Discogs was an impossible mess on that one as they take the cover text - complete with error - as more important than the actual lengths.
The biggest problem with a track time on a vinyl is always a choice of the user as to where to split. Or the written track list not actually including the chat between tracks.
My real question here is - did the CD release add anything? If nothing was added, then I would try to keep the recordings the same. If nothing was added to the vinyl version then it is technically possible to apply the times from the CD. Just visually it would likely look better with the times from the cover artwork.
If downloads exist of rips of the LP cut in a different way, then I tend to lean to making different recordings to match this. I’ve seen it happen with music in my own collection where a hard to find vinyl had been cut up in a specific way in an easy to find torrent. Or a download from the artists site. These times had been applied to one of the vinyl listed in MB. Along side it was a vinyl with times as written on the cover art. And then an official CD came out with a new set of ways to split the tracks. This has led to separated recordings for some tracks as it makes more sense.
I don’t really like deleting track times just because I can’t tell if they are correct. I’d like to trust an editor.
It didn’t add anything, but the different times are associated with different recordings based on the spoken content between tracks.
I started working on the Reprise releases, of which there are 3 in MusicBrainz. Two of those three (R 6199 and SR 6199, the ones with track lengths that currently match the CD) started from a freedb import in 2002. There have been subsequent updates to recordings from all manner of sources, but almost all updates to the releases themselves have been bot edits (a bot added the discogs link, later bot updated details based on discogs, etc). I’ve initiated an update to R 6199 based on the attached discogs, and can’t find any evidence of an SR 6199, only RS, so preparing to merge those.
I hadn’t looked at the two earlier (Decca) releases in depth yet, but a quick glance shows they are tied to the same freedb import in 2002. There were at some point DiscIDs in the mix (1, 2), which are now attached to the CD. That would explain why all the lengths across these releases match.
I would be tempted to just keep it as one set of recordings, just add a note that the vinyl shows different written lengths due to where the voice sections are taken from.
Technically the vinyl can be split in the same way as the CD. So the times written on vinyl covers are just hand written splits.
OR just do a new set of recordings for all the vinyl and disambigs to say “starts with chat” or “ends with chat” on each variation.
I remember a conversation about a Pink Floyd album where the exact same audio is on the USA vinyl but someone wrote all the wrong lengths on the Vinyl label. I think in those cases the times got adjusted to actually what you hear and the artwork completely ignored. Annotation then noted the reality.
Discogs can often be a bad source as they blindly stick to the text. I think they are more aimed at putting the music on a shelf to sell, so write what they see. Whereas Musicbrainz is more about what you hear so write the lengths of the actual music as it is played.
So started from a CD rip… and then later converted to link to vinyl.
I think the key is as long as you do careful focused edits and note the differences then your work will be an improvement over a mish-mash of partial edits.
With the vinyl (short of, as you say, a release in hand, which doesn’t seem to be the case for any of these), all we have to go by is the printed times. There are, however, two documented recordings for these specific tracks, which can be found on CD and digital releases (not counting edits on compilations - have found at least one of those, too).
So the vinyl could be one or the other or totally separate, but I’m inclined to put it with the recordings that match the printed times. (To your point about covers, the times printed on the CD match neither the times given on the vinyl releases nor the actual times on the CD according to 3 disc IDs.)
Another check I do is look at the lengths of the AcoustIDs. Digital often comes from the CDs. Especially with older stuff like this.
If you end up with two clean sets of recordings - one for vinyl, the other for CD/Digital then that it is clarity. Then if someone rips their vinyl to have the tracks match the written times they have something to line up to.
Just to muddy the waters, some versions of these two sets of recordings have overlapping times. In fact, in one case the changes in the two versions balance out (removing from the front end, adding to the back end). The only way to know which one you have is by ear or by looking at the other recordings to see how they line up.
I’ve seen that kind of overlapping thing happen a lot. Sometimes pops up on remastered CDs where they swap the chat from the end of one track to the start of another.
Or even more bizarre where you find a track where an intro has been accidentally left on the end of the previous track. (had that last month)
All a trick of matching up pairs of lengths. I recognise that game.
(I haven’t, at this stage, tried to account for stereo vs mono recordings. Those were combined when I started, so if they get sorted out it will be a later step.)
If you ask me, I think removing times a bit too much. Just changing them into the values you see on the cover is enough. Or retyping them without the decimal places of the discID.
But I suppose removing them nudges someone with a copy to actually time the tracks.
There are no values on the cover for these releases. Using the CD times causes confusion around identifying the different recordings, so removing them falls under “no data is better than bad data.”