ISRC duplicates

Examples of ISRC errors

I dived into my old edit history and found two examples/

This one was a guess that it was wrong as same ISRC attached to two tracks on same album. Leading to one track with two ISRCs.
https://musicbrainz.org/edit/88613051

This is a good example of reusing ISRCs on very different sounding audio:
https://musicbrainz.org/edit/76827998

Sorry for the rough opinions from me, but I am not a fan of the money side of music. I prefer to listen. But read past that to the details. In there you see examples discussed where different Recordings from different Pink Floyd albums are showing up with the same ISRCs. These are reissues, remixes and other places using the same track.

You see my notes in that edit when I put these albums on and listen to the tracks and explain clear differences.

Fearless originally appeared on Meddle. Then appeared as a different edit on the compilation Works (where it is audibly mixed into the following track “Brain Damage”). Then again as part of the box set “A foot in the door”. You would treat these as two distinct recordings if you listen to them, but for financial reasons they share the same ISRC.

Then look at the Fearless page linked and you see three ISRCs on there anyway. Some of these are likely to be reissues, anniversaries, remasters, etc. EMI and boxset \ reissues build confusion. EMI is especially bad at ignoring the uniqueness of ISRCs as they are money first, music second. There are a lot of EMI Cash-in reissues with little differences from the originals beyond a new box.

With just this one track we see EMI reusing an ISRC on what should be separate Recordings. And issuing new ISRCs in boxsets on what is actually identical to the past.

1 Like

I still think the best way to track ISRC would be to have them linked to both recordings and 0 or more tracks. [MBS-9754] Make it possible to add ISRCs to tracks - MetaBrainz JIRA asks for that, and we had this discussion before I think.

Then it would be easier to clean up and actually see whether a single recording indeed has multiple ISRCs or if it was because two separate recordings got mixed up. In combination with other tickets that ask for editing ISRCs for an entire release this would be convenient.

It still would need to be optional, though, and it must still be possible to associate an ISRC with only a recording. And existing submission tools would need to get updated.

2 Likes

Seen a few of those cases. Happens with compilations and best-of albums. A clear indication that something is off is when all tracks on a compilation have consecitive ISRCs, even though it can be clearly shown that it contains the original album recordings. This usually clearly shows the label did just assign fresh ISRCs to all tracks and didn’t bother looking up the actual ISRCs.

1 Like

Thanks for the examples. I will look at them further so I can see the full picture of what is happening. I do want to add… those are both older releases. I am not discounting them by any means, but these fall into what I mentioned above, the historical issue. It is good to see these however, so that issue can be further understood if at all possible.

Based on the above discussion, different sounding audio is a result of mastering, which is not relevant to MB in reference to a recording. So, the reused ISRC would be just fine, as the recording will not change anyway… Correct? If I have a recording (a mix which is a MB recording) and I make one copy on cassette and one copy on 24 bit FLAC, most will be able to hear a difference. However, that difference is a result of mastering, not mixing, so it is the same recording. Please correct me if I have misunderstood.

In this example, the different sounding audio has other tracks mixed into it. Works is a compilation and has one track fade into the next with some very clear overlaps. These are separate recordings in MB eyes (and I have separated them with different disambigs). These share ISRCs in EMI’s eyes.

But also there are remasters of these albums that have been released at various occasions. These get new ISRCs from EMI, but as they are “only remasters” in MB eyes those Recordings are merged. This happens with multiple editions of a whole album like Meddle.

LOL! I do a lot of “old”. Majority of my CDs are 1990s :rofl:

This is good. Linking to tracks keeps them attached to the albums. Handy to see when they are talking of remastered versions, etc in the above Meddle examples.

1 Like

I remember a few of those as well, but cannot recall the exact ones. It is my belief that these are older releases. Meaning that either the comp release or the recordings used on it are older. Older meaning pre-internet age.

It is my belief / assumption that this issue was far larger when there was no easy way to connect the data between locations and companies. While this is an assumption, it seems to be supported by the data. Artists that are older have more duplicates and artists that are new are less likely, and most often, I see they have none. I could be wrong, this is an assumption based on my observations.

The reason I mention the age of the release/recording is because my use of the ISRC is for digital releases mostly. The older releases, say a CD, I can use the barcode with catalog number to identify. With a digital release, those items are not relevant in many cases. If that information is not in the metadata, it does not work for the purpose of identification. The ISRC however appears in almost every digital release I have. This I understands depends completely on the source of the release. I am only explaining my logic. I know that digital releases from streaming sources are different with all that as metadata is not really relevant, but personally I do not consider streaming sources to be the same type of release, and I personally disregard them for my purposes. That does not mean I suggest they be ignored, it just means that is what I do. Others likely use them for something.

Not with Pink Floyd. With all the legal battles over the years there is a very clear trail of what happened and when.

Also note with other bands that many compilations are made from a label just diving into a catalogue and pulling out tracks recorded long ago. They know it was only recorded once, but attach a new set of ISRCs for the compilation. You can spot it happening when you get a compilation with sequential numbers.

Yes, I see this as a problem in many other areas as well… if anyone treats mastering as different, it will cause an issue in MB as mastering is mostly disregarded. I personally disagree and place higher value on mastering, but we work with what is here.

That makes sense. I assume that to be the case with others like Prince, where the legalities and royalties are under heavy discussion. I bet they have teams of people gathering each and every ISRC for each and everything ever created.

I might also assume that any artist/band of that high of reputation might also be the same… The Beatles, Led Zeppelin, etc.

Looking at the Nirvana example, release Bleach and recording Blew, there are two ISRCs, US-SUB-02-00002 and US-SUB-09-83401. This seems to support my theory as when I look up each of those ISRCs, there is no detail added in the enter. No release name, no label, no duration, no year, etc. As someone mentioned above, this is a problem as there is no way for people to differentiate them. This again seems to be an older issue though, as most all modern ISRCs include all of the detail. It is interesting to see this scenario though… there is really nothing we can do about it.

1 Like

ISRCs will only be a small part of that legal paper trail. Which is why I think MB should also mimic. The number is important, but not unique.

These new ISRCs can especially be seen when someone like EMI use an excuse of releasing a set of “digital remasters” as a time to refresh ISRCs so as to restart a copyright date. Sometimes the list of (p) dates on a recording gets comically long.

And yes, the ages of some of these bands is very relevant as ISRCs were hard to embed in vinyl.

If you glance down the RG page of Meddle you can see the 1971 vinyl, original 1980s CDs, 1994 Digital Remasters, 2011 Discovery Remasters, 2016 Another Cashin editions… Many getting new ISRCs from EMI, but the amount of actual remastering happing now getting more and more questionable.

Some of the problem with the Bleach example appeared when I looked at the Release page of Bleach and saw the same ISRC duplicated on the first two tracks. I took a guess and deleted the one that looked most wrong numerically, but trying to locate where it was added from was tricky. If ISRCs were attached to Tracks as well as Recordings the trail would be easier to check.

This is a very interesting topic. While it is centered on older releases, which I suggest is the problem area in my statements, it is still very relevant. Please note that I have not done any of the edits I mentioned on such older releases, it is modern only. I find it interesting as I also place a good deal of value to the mastering. However, to your statement, I really only give that value to the remasters that actually sound different. I agree that at a certain point, the remastering becomes irrelevant when the difference can only be seen electronically. The logic I use is … did the remastering actually provide a value to the end user or not.

Thanks for further explaining. For me, this is why I avoid those releases and issues. When I see even modern ones there there is, for example, a mixing of ISRCs on recordings of clean, explicit and not stated, I just leave it alone. I order to properly change it, I would need to have the relevant releases in my hand so the ISRC on the actual release can be seen. Especially when it is not stated as explicit or clean, assumptions can only lead to problems and confusion. I understand there is a portion my small initiative cannot fix, so I settle for what I can fix and leave the others. I would prefer to fix on facts vs assumptions.

I am interested to see if anyone might find a way to address those issues in MB. I fear the lack of attention to mastering in the database might prevent this though.

I can hear differences in some of those earlier remasters, but does 2011 sound different to 2016? Is there really much change from 1994 to 2011? It is why I generally started seeing the point in MB’s “just mush them all together as it is a gimmick to sell albums”.

I picked a simple one with Meddle. If I had picked Dark Side of the Moon it gets even more complex as you then add in 20th, 30th anniversary editions. I know I have my preferred copy I pull from the shelf, but if I was a true addict then I’d be chasing down specific Japanese pressings from certain factories. This is not a distinction that appears in ISRC numbers. This appears from packaging and matrix details. ISRC does not give enough detail for finding those. “best” recordings will always require a combination of data to locate.

I guess some of this is different for me as I look at whole albums, and not loose tracks.

I wanted to share this artist,

If you look at the recordings, the ISRCs are all clean, no duplicates and one per. This is a newer artist though, but one of reasonable popularity. I wanted to provide an example of my side of it, to show the opposite of the examples provided to me.

She is fairly simple. Not many albums, limited reissues, and no Greatest Hits compilations. Only been around a few years so not had time to drop into those “back catalogue VA compilations” from the labels.

I understand when it is perfect and neat, a bit like AcoustIDs, but real world data gets messy. :slight_smile:

We aim for perfection, but also deal with the messy stuff along the way.

1 Like

Absolutely. It will be interesting to see how such artists develop with the ISRC issue. I for sure see the issue and how it relates to comp releases.

Some individual record labels might do that, but it’s completely up to the discretion of the record label or ISRC issuer. The fields in an ISRC are all pretty simple - 2 character country code, 3 character registrant code, 2 digit year, 5 digit designation code assigned at the discretion of the issuer.

I’ve seen some labels that use the first 4 digits of the designation code to identify a recording, then the last digit is 0 for original, and incremented for edits. But I’ve also seen the full 5 digit number range being assigned sequentially, or out of blocks.
There’s no way to derive any meaning from the designation code numbers without knowing the details of how a particular record label assigns numbers.

This was probably a bug with a CD drive or software. There’s a few cases where the drive/software can get out of sync, and repeat an ISRC when reading the info off the disc (in some of these cases, it results in some or all of the following ISRCs being on the wrong track!). I put in a warning on duplicate ISRCs on https://magicisrc.kepstin.ca/ to help prevent submissions like that.

8 Likes

That’s actually one of the reasons why the ISRC submission was not yet added to Picard (the other reason being that simply no one has yet started doing the work).

1 Like

Example where more than one ISRC for one recordings seems correct (year 2015):

released by three different labels - here it’s just about payment…
Germany: Here It Comes - Album by Life in Film | Spotify
UK: Here It Comes - Album by Life in Film | Spotify
US: Here It Comes - Album by Life in Film | Spotify

Example where two recordings sharing the same ISRC (year 2008, both digital releases):
https://musicbrainz.org/isrc/FR0PO0800050
Last Night I Got a Train to Moscow: DYLB - Single by The Dodoz | Spotify
Stanislas: The Dodoz - Compilation by The Dodoz | Spotify

And yet I’m a fan of ISRC as an identifier. Especially as they help preventing merges of different edits/versions just because they share the same track time.