Found a recording with different recording dates

Found https://musicbrainz.org/recording/ea24b949-b539-452a-9249-3324c8556a1e with multiple recording dates, most likely due to a bad merge.

If it can be safely determined which releases to assign the split recordings to to:

  • I’m guessing that the best thing to do is to split the recording, I’m not sure if it is best practice to create two new recordings and discard the original or if I should retain the original recording
  • Is there a way to determine and reassign the acoustic ids, ISRC etc… to the new recordings?
  • should I notify the user’s who previously merged to bring them into the discussion?
3 Likes

Based on https://isrcsearch.ifpi.org it seems that releases 442 589-2, 462 699-2 & 475 7160 are having a recording with ISRC NLA506800336 but 411 137-2 is having NLA508301814. Recording year (with Philips recordings) is usually the same as 2 digits after NLA50 on ISRC.

I would split it to 2 new recordings. ISRC can easily be added or removed (recording editor has a field for it) but acoustids can’t be moved (but with Picard can be added again). I usually contact other editors only if I’m not able to fix the mistake by myself. I wouldn’t mind if being contacted when there’s any need for help.

1 Like

Just want to add that info from ISRCs should be treated with care. I have seen it too often that labels were just reassigning new ISRCs for exactly the same recording on re-released songs, especially when they are released as part of a compilation / best-of

1 Like

I agree that this rule doesn’t apply to all labels. That’s the reason why I especially mention Philips with my reply. Some labels tend to keep using same ISRC for all re-releases but for example Sony seems to create new ISRC for every re-master. ISRC should never be used as an only source for updating dates for relationships because it could include either recording year, remastering year or release year. It still can be often used to identify correct recordings on situations like on this topic.

1 Like