Different ISRC or masters (not) sharing the same recordings

Tags: #<Tag:0x00007fd5e79a11a8> #<Tag:0x00007fd5e79a0eb0> #<Tag:0x00007fd5e79a0bb8>


The “Cool Kids” ISRC listing is a nice clean example. Here’s a counterexample, a fairly obscure Slim & Slam track, “Ti-Pi-Tin”. I picked it because I’m reasonably certain there was only one original recording (on Feb. 17, 1938). I’m very skeptical that all 7 of those ISRCs represent unique restorations/remasterings of the original. In a case like this, I wouldn’t want to infer anything based on a different ISRC.


That makes sense to me. It’s also probably safer to assume that an unknown release doesn’t use that specific 1996 remastering than that it does. Okay, I will leave my merge to apply, and then your edits can apply a day later, and then you and I can review them to make sure we’re both happy with them.


You are absolutely correct, there are many recordings that do have different ISRC number assigned where they may not have been needed. If you look at MB though (https://beta.musicbrainz.org/artist/b38c27ca-c76c-41e2-aaa2-c4a84aa06b86/recordings?page=3), that is about how many recordings of that one there are anyway. YOu also need to consider that on the list you provided, artist credits are different as well as countries, so for sure if you want the actual differences, would need to look up those releases they are used on and see.

I had used the Cool Kids example asI just fixed a listing that mixed up the two versions. It was just chance that it was clean. Also, it is a newer recording. I would assume that older recordings are not so clean.

My point though is that no system will be perfect, as people will use the system, and people are inherently faulty. But with the ISRC, if you have that number for your recording, it is far easier to identify what the recording is, meaning mostly placing it on a specific release to identify the exact recording.

I also like the way I can look at my recording and easily get all the releases it is on, that exact recording… same sound, same acoustID, etc. This per the above obviously more relevant to newer releases than older ones. The reference you provided does not even list exact releases, UPCs, etc. The data is just incomplete for those years.


I would tend to agree, if all others do as well. Let it go as the resistance to my edit has relaxed (unless someone else wants to vote it down) and look at it after to see. Worst case, we can merge the recordings all back in.

I think though, as a general guideline, if a person can uniquely identify a set of recordings with ISRC, specific release, acousID and the verbiage that describes the difference, there would be no harm in making separate recordings as you can make them very specific. I just added a Paul Simon 2CD release and I was not happy. The recordings all have multiple ISRCs and a ton of acoustIDs. Just from having the release, I can see that all that data is not correct. So my only option is to create all new recordings or just use what is there. I used what is there, which improperly represents the release unfortunately.


I am sorry this is starting stray from the intial intent of this post, but this thought has come into my mind in this discussion…

So, in general, MB does not care about remasters. So given this, what about masters? To be very clear on what I mean… a remaster is going back to a prior release and mastering it again, whether a full remaster or a quick remaster. So that aside, when I make a new release, I have a master for different purposes. This gets into a very complex area. One example… the master used to make a standard iTunes release is different from the master used for a ‘Mastered for iTunes’ labeled release. The difference being that mastered for iTunes is required to be a 24 bit master and is desired at 96khz, although a 44.1 can be accepted… but 24 bit is a must.

Where this is going, in effort to keep it short, is do different masters matter? this will be almost direct tie into stores, file types, container types and quality. Very generally speaking, you have the FLACs (like HDTracks), the mastered for iTunes product (similar to a made from HDTracks in a sense), Amazon MP3s (from 44.1 16 bit masters) and CDs from the same 16/44.1 That is just to name a few of them, there is no intention on excluding anyone here, I just speak to what I know.

I would guess that since a remaster does not matter, that different masters do not matter either. This would imply that MB will not be differentiating between digital releases, meaning that a 128 MP3 will be the same as a HD FLAC as far as the recording is concerned. [I am not touching release differences here, this is only recordings]. Given the above discussion, would this be correct? The reason I am asking is on the same topic of sound and acoustID. There is mention that if there is a clear sound difference, that can be a new recording. I would hope that all will hear a difference of the two files named above, a 128 MP3 and a HD FLAC. At the same time, these two may very well have the same ISRC, possibly a different acoustID and sound different… which could lead to 2 different recordings with the same ISRC.


For those following this conversation, I wanted to add an edit from a release I was just editing dealing with a mastering issue.


On this release, RE that edit, there are different versions of this release, distinguishable by the mastering and other factors not listed on releases here. I am just curious… if I were to try and clean them up and attempt to identify which is which, how would I do so?


The matrix of the CD is probably different (different factory, different mould, etc.) at least one difference.
It’s better to differentiate masters on releases because when it was on recording level, we would have one new recording for each release because it’s impossible to know.


@jesus2099 - ahh, ok… so I can add mastering data onto the release. I thought someone had said those fields / relationships were obsolete. I may have misunderstood.


It’s only deprecated at recording level - you’re more than welcome to add that info on the release :slight_smile: A few times there’s different mastering info for different tracks - for now, the best option we have for those is to just specify in the annotation what tracks stuff applies to.


Ok, I misunderstood. So recordings = no mastering data. Releases = mastering data ok.

@reosarevok - could you look at the post right above this one that jesus2099 replied to, in reference to mastering and digital releases? I wonder how this fits in.


I have been following this topic from the start and tried to stay out, but am compelled to make some observations without taking any sides. The way ISRC’s have been merged (up to this time) and the destructive nature of those merges means very little to nothing can be done to revert them, the “source context” of that data is lost, the only context left is the recording. Whether that’s good or bad depends on your point of view and what you need from the database. I point this out because of the desire to add “contextually accurate” data into a “system” where the existing data may not be “contextually accurate”.

I noticed this merging several weeks ago when I had to manual add ISRC’s to a album instead of using idtcsubmit.py due to damage on the medium. Further searches showed much more of this merging to the point I almost posted a question about it. I decided it was done for a “reason” and since I keep all the information that is available from a rip (including logs) I still have the original data.

IMHO the practice of slapping a different ISRC on the same recording happens mostly on compilations produced from a different label/corporation than the originator. I stopped buying Rolling Stones compilations because according to the ISRC’s I had 5 copies of the same version of “Jumpin’ Jack Flash” (that’s an example). I have found that the same recording released from a different country US vs. GB may have a different ISRC. As I am writing this is see my US copy of “Shine a Light” has all GB ISRC’s.

What I am trying to point out is what others have said, ISRC’s may get placed on recordings for reasons we may not know or be sure of, the labels and labels/country do “strange things”, and those “strange things” may change over the years. I will say the present practice of merging ISRC’s is a loss of data and I am not yet sure what that means.

Database design is not simple and trade-offs may be made for performance reasons. I would have to analyze the schema in depth to better understand the impact of many recording/ISRC’s to fewer recording/ISRC’s where relationships exist.

IMHO I do not think using the ISRC to differentiate recordings is accurate or ever could be accurate, ISRC generation is at the whim of the label and we have no control on what they do, when they do, or how they do it. Even the use of the ISRC’s in the database for royalty payment cannot be guaranteed.


Thanks for replying and explaining your observations. I think you have pointed out great points. I do not think any system will ever be perfect. I also want to say I appreciate your more neutral approach. The only reason I have a fairly strong side is one side is a guarantee of data duplication, while the other side is guaranteeing data loss/destruction. I do not mean to make that sound large, as the actual result of that is minimal on both sides.

Question for you on the “Shine a Light” release. You stated that your US copy has all GB ISRCs. I am unsure of where you are going with that, fully. It is my understanding that the country of release is not to matter, but the country of origin. I may be using the wong terms there, but I understand it that once the recordings are issued an ISRC, in lets say US, that same ISRC is to be used in all other countries it is released in.

On further research into this matter since starting this. I agree that ISRC are at the option of the “owner”, and we are likely never going to know each and every reason for it. Some might be clear enough, like a remix, and others may appear to have no difference. That we can never change. On the other side of that, we also cannot change when a release is printed with incomplete recording data. Like the “Cool KIds” example I stated, the release originally used “Cool KIds”, and was reissued with “Cool Kids (Radio Edit)”, but the release kept the title as “Cool Kids”. I have a large number of examples of this, and the only way you can tell is in some cases the duration is different enough, using the difference of ISRC or having the knowledge of the issue.

It is for those reasons I personally like the ISRC. I would rather have a safe than sorry practice, which MB has anyway indicating that you should keep recordings separate and not merge unless you are sure they are the same, to be safe and not sorry. This will knowingly create duplicates. On the other hand, combining data also results in examples I have found on which MB lists recordings for a release showing the wrong ISRC codes, meaning not the ones used on the release. To me, that is more destructive as the release is misrepresented by an assumption of someone who “thinks” they are the same.

I do not personally care much about works and different releases in terms of packaging. This data is important, please do not mistake MY personal interests vs what should be in MB. I care about the recordings on the CD or in the files. If a CD has all of the same recordings, but one is a GB and the other a US release, for my collection, I do not care. I do however have the data on where they came from, so this makes the MB additional data great to have. I personally differentiate my collection on real differences in the music, so if a remaster for example reduces the dynamic range and makes it hot, that is absolutely a set of different recordings, because they are not interchangeable in use. Whereas a recordings that sounds the same but is on a US and a GB with different markings and printing locations, I do not care as the end product (the music) is no different.

But I still support the MB concept of identifying releases. I think capturing that data is at very least nice to have. But I also think that if one is to make all of these distinct (and correct) different releases, they should be properly represented based on fact, not assumption. I think it is nice to see the different releases of a CD, and I would love to see the different releases of digital as well, but I would also like to see within these different releases, what is the difference… in package, recordings, mixes, etc. If I purchase two albums on the list, which recordings within them are REALLY duplicates vs different in some way?


I used the “Shine a Light” example to show that what we expect may not be accurate. More examination of the package reveals a generic country-less package except for the outer cardboard sleeve over the jewel case which has the FBI logo which in theory makes it US, but the overall package smells worldwide, with maybe mastering in the UK. It would require more investigation than I am willing to put in. But it is a good example of the labels screwing with us.

My only point on the ISRC’s is the merges have been going on for a good period of time. IMHO the present merging of the ISRC’s is not reversible. It could be stopped, but for what purpose? For me I am not worried since I have the data and I am finding more and more their value to me is suspect.

I could make comments about release dates, but I won’t.

The ISRC’s bother me much less than the discid’s which are applied to a release by a human making a judgement call (or maybe not thinking at all), but that’s a discussion for some other time.

All said this is the best database out there and it keeps getting better.


I see your point. At least the ISRCs were not made new (I assume based on your statement), which would be a good thing as that is how it should be. I would guess the different barcode on the outside is just allowing te different country data for sales to be there without changing the CD and the process of manufacturing it. I personally see no confusion in that, but I can see how some would, especially those who collect multiple issuings of releases, or want a certain one, etc. you are then thinking you are buying a US release and end up with a US rebranded UK release.

The only real thing I disagree with is the concept of if you see something broken, just ignore it. I do agree that the merging of ISRCs cannot be undone, at least easily and within reason. This is very true. But I ask why keep the “problem” spreading if we agree it is in fact a problem? I do agree and do the same as you… to some degree I am not bothered by bad/incomplete/etc data here as I have the proper data here with me. My only “issue” is being detailed, so I see issues like this and want to fix it. For a purpose as you asked, there is a correctness it can offer that is currently not there. The example I mentioned prior of a recording on a release being swapped and not titled any differently on the release is just one example.

The only reason I am looking at ISRCs here now and prior in MB I was not is because I am getting more and more releases coming with clear ISRC identification. So this caused me to look into them further, thus starting to include them in my edits here. Upon doing so, I found that recordings are combined improperly here. I pass no blame, as I guarantee I have made some of those edits myself before becoming aware. This was great for me, since many new releases are heavy digital, some not even having a physical version. On digital, barcodes are more passive and sometimes not even disclosed. This made the little identification we had and made it harder. Then, all the digital files starting coming with ISRC data in the meta, then bam…means of identification was respored, and from my perspective within my uses, restored better than it was before. This is also realizing that ISRC data until recently was not so easy to get. Now it is reversed, most digital files now have them giving us ISRC data on both digital and CD formats.


Just now in my editing, I found this recording with 2 ISRC numbers assigned to it:

Currently in MB, this is fine, as one is the original and the other is the remastered version. The reason this does/can make a difference is what if I am looking for what releases use the remastered version? I know the data here, so I can look for releases at and after 2007 and partially narrow it down, but the date does not mean they did or did not use the remaster. All I can for sure say is that if it is prior to 2007, the remaster is not on those releases.


I’d call it 1996 remix instead, because guidelines state that remasters should be merge with original recordings. If you truly believe that it’s a significant change from the original, than isn’t that a remix?


I see you point very well. However, if I call them “96 remix” vs “96 remaster”, and someone else looks at it, they may change or otherwise edit because the recordings are not marked or titled “96 remix”, but “96 remaster”. This is my dilemma. If I mislabel them to prevent unwanted merging, I could open the door to unwanted editing as one could see those recordings as wrong, etc.

I am open to what you guys (and gals) think. My intention here was to properly represent the release. But if I need to improperly label recordings to do so, I am not really properly representing it again. I can clearly see why and how this type of stuff does not and almost cannot fit in here.

I have the 86 and the 96 copies of this release. What I can say is that the recordings (although the listing is slightly different) are not interchangeable. But maybe in MB they need to be. I think it is important to properly identify the music you are looking at/for/etc. If I want a CD, vinyl listings are not suited to me. If I want a FLAC, MP3 listings are not suited… and that last example is somewhat equiv to a remaster such as this, same “song”, but they do not sound the same and cannot always be used for the same.


This is the key “Digitally remastered from original metal parts and mastertapes”. This means that someone went through the hassle of locating the original sources vs using safety copies or multi generation copies to do the master. Now was there a remix? I cannot say, only assume/guess. But this tells me that the source used was first generation.

EDIT: I should have said … “This SHOULD mean that someone went through the hassle of locating…”. I cannot stop or have any way of knowing the level of honesty, the the quality does support what it says.


I agree with thwaller that the 1996 remasters are audibly distinct from the older versions. In particular, there is noticeably less background noise.

I would not call it a “remix” myself. I don’t think I’ve ever even heard the term remix applied to recordings from before the multitrack era.

If it were up to me, I’d apply a sort of “reasonable person” standard. Is the difference between two recordings noticable to an average listener (as opposed to an audio engineer or audiophile listening on a high-end system)?


Here is a question on this topic, given the core answer is on the table…

I just did some editing of this release. In the process, I decided to look at the ISRCs assigned to the recordings used. I only checked the first few and stopped because each one I checked did not have the proper ISRC in the list, so all of them listed on the recording are “wrong” for this release.

So right to the point… this is ok with those here? To me, I could add the proper ones, but I am then adding an ISRC to a recording used in many other places that obviously does not fit in… or I could not add the proper ISRCs and have the release misrepresented.

Some thoughts of mine to add to the (hopefully) thoughts of others… I can go through and check each of the ISRCs to those used on the release to see what and if there are differences. Then listen to each of the recordings to see if there are audible differences, or if nothing can be heard which would justify them being all merged. Now since we are dealing with approx 20 recordings, this is a whole lot of work to expect on someone wanting to add/modify/verify a release. I could do this as I have this release and most likely a good selections of the other ones in the mix using these recordings. But how many people can realistically do that? So my options are to rely on different ISRCs and say these are different, pay no attention to the ISRCs and call them the same since the title and durations match, loook at acoustIDs which is no help since I cannot compare the IDs on my recordings vs the IDs listed in MB… or I could just say not worth it and not bother verifying anything, or in the case of a new add, just creating all new and having someone followup and do mass merges.

What is the recommended course of action for an editor who is average and likely does not know all of such detail and also likely does not have a slightly short Curtis Mayfield discography? I have left the ISRCs and recordings alone at this time.