Different ISRC or masters (not) sharing the same recordings

Tags: #<Tag:0x00007f4d55a09980> #<Tag:0x00007f4d55a09868> #<Tag:0x00007f4d55a09700>


I have an edit in which I am creating new recordings for a release so I can add the proper ISRCs to them. Because the recordings on this release are used on other releases and some have different ISRCs already assigned, I need new recordings to do this so I do not assign ISRCs to recordings used on other releases. These recordings are the 1996 remaster versions. Because the edit needs to pass vote, I have to wait a week before I can finish the edit, add the ISRCs to the recordings since they will not be created until my edit passes.

Should any agree with what I am doing, could you kindly vote yes to pass this sooner than later, or an auto-editor help pass it?


Does a remaster imply a new recording?
https://musicbrainz.org/doc/Recording seems to me to say it doesn’t.

1 Like

The logic I am using on the edit is the ISRC: https://beta.musicbrainz.org/doc/ISRC

“assign a unique identifier to every distinct sound recording” and
"An ISRC identifies a particular sound recording, not the song itself. Therefore, different recordings, edits, remixes and remasters of the same song will each be assigned their own ISRC"

So my thought process tells me that since the record company, label or the party(s) involved have considered this set of recordings different and unique in comparison to other versions/masters/etc of the song. So I am not using the fact that they are remasters as a sole factor, but only a partial factor. All factors including acoustID, ISRC, edit/mix/etc, release notes, etc.

Additionally, Discogs shows a version of this release marked remaster and one not marked remaster. This data may or may not be correct, but the information I gain from such a thing is there appears to be some differentiation going on. Also, when you look at the track listing, not all of the recordings are marked 1996 remaster. There are three that are not marked as such. So my intent is to make the recordings used match the ISRC listing.

For what my opinion is worth… I disagree with this reference. A remaster is different and should be treated different in most cases. I would like to reference this as supporting data: https://en.wikipedia.org/wiki/Loudness_war

In specific, please note this listing:
Nirvana Nevermind (20th anniversary remaster)
Now this remaster is referenced and known to be poor to the criteria of the reference link. What makes this 20th anniversary remaster different from the original? Remastering. The mastering process that was used on the original produced a different sounding release and recordings than did the mastering of the remaster. To call them the same would be wrong in my opinion as they sound different. I do not see how you could call two recordings that provide a different listening experience the same when we remove the differences created by the medium itself and keep a like comparison. If remastering/mastering made no difference on the musical product, there would be no point in doing it and advertising special releases noting that it is a remaster. Normally the end user is interested in a remaster as it should provide a better listening experience.


I see there is a “remaster of” recording relationship available, which would support your position over that of the reference.

1 Like

I see there is a “remaster of” recording relationship available, which would support your position over that of the reference.

But when one tries to add this relationship to a recording, one is told in no uncertain terms: “This relationship type is deprecated and should not be used.”

From https://wiki.musicbrainz.org/Style/Recording:

In Music Brainz, mastering is a process that is applied to recordings to prepare them for release in a particular format. This means that tracks should not use separate recordings because of mastering differences.

Following on from this, separate recordings should not be created for remastered tracks, since remastered tracks generally feature the original recording with different mastering applied. Remastering should be described using the Remaster Relationship Type between releases, or in the release annotation where tracks are mastered differently across a release. The exception to this is where a track labelled as a remaster is in fact a remix - in this case, follow the remix guidelines above.


If you look at this conversation I just had:

The mastering process creates a very different looking fingerprint. Since editors reject merges based on fingerprints, there seems to be a direct conflict here. If a fingerprint mismatch is reason not to merge, then a master is a reason not to merge.

So basically, you are telling me that a fingerprint is not to be a factor in the merge or non merge of recordings?

So the second part is telling us to use the remaster relationship that the first part is saying is depreciated?
Second part: Remastering should be described using the Remaster Relationship Type
First part: But when one tries to add this relationship to a recording, one is told in no uncertain terms: “This relationship type is deprecated and should not be used.”

or are they referring to different things?

I think it’s saying to use the release-release relationship rather than recording-recording.


Copying over one of my comments from the edit notes for wider discussion:

For my part, I consider performer relationships important, so while I don’t have an issue with your notion of creating new recordings, I would not want to lose those relationships that are now available (indirectly via the track level) on release https://musicbrainz.org/release/33729c11-dad9-4e… … If that doesn’t happen automatically (and I think you’re right that it doesn’t) then I’d say it’s your responsibility to add them to the new recordings. Otherwise you’re improving the release for yourself (adding information that’s important to you) but worse for others like myself who care more about the performers than the mastering.

Rereading my comment, I want to clarify that I don’t mean to accuse thwaller of being selfish; he’s simply trying to apply a consistent policy regarding ISRCs.

I, on the other hand, am being selfish, as I care about performer relationships (particularly for tagging purposes) much more than I care about ISRCs. :grin:

1 Like

You are all good, no worries.

I think all the data is important. I would prefer to lose the performing artists in favor of the proper ISRC for the main reason that I prefer partial correct data rather than complete incorrect data. I of course an generalizing with that statement, but to combine 2 recordings that have different attributes is wrong to me.

For me, and this is just my personal opinion, I do not tag with performing artists or other more detailed credits like that. I am finding it best to tag with items like a UPC, ISRC, etc. If I have that info, I can then easily find the rest in other places. I believe that the most important factor in tagging is to properly identify the song/recording/etc, what ever you want to call it, that the file contains. For proper identification of performing artists, ISRC is a great tool, since a change in performing artists warrants a new ISRC. I am fairly new to using ISRC in such a way, but it was motivated by the many difficulties of tagging in addition to iTunes and other stores listing this for all the recordings sold in their service. So it is becoming a constant that all recordings tagged will have.

I do not tag any longer with Picard, but for tagging, I know you can select the release you want to tag from. So if the release you are tagging has 5 versions, you can select the one to use from the MB database, thus selecting the one you have filled out completely. A different release could very easily have different attributes, thus why it is a different release in the first place… like in this case, the credits relating for mastering, the acoustID, the ISRC, etc.

I believe it is the case for all recording with multiple versions… the extra relationships like performers have to be entered to each. Maybe the classical people can assist on how to handle this better as there is a lot of detail there. I think having the performer information within MB is great, and although I do not want it in my tags, it is a place I can get that info from the data I do tag.

Briefly, since it is sort of off topic but mentioned… I have struggled with tagging and settled on the idea that tagging with the common data of artist, title, album, album artist, date, UPC, etc…, having ISRCs for each track, etc. makes it easier. Reason is that it keeps the tag data smaller and it provides me with all I need to accurately identify the release and each of its recordings. So I can take my reference data and be able to pull any data I want for that release and its recordings, assuming I have a place to get such info… like MB. I am hoping that other large sources like iTunes, Spotify, etc improve on making data available as they are all now using ISRC and UPC as identifiers, which is helping create a common ground.

Another example of this in the community:

RE the specification of remastering.

Not many are commenting here, so I am just adding current examples of this issue and what it causes, at least from my view as an editor making edits…

PLease see this recording:

It is assigned to 2 ISRCs, one of them being a remaster. Now, I came across this adding a new release and saw a crazy number of acoustIDs on it. Looking at why this could be, I found the 2 ISRCs on it, and one being remaster, just like this case here. To me this is somewhat problematic if the acoustID/fingerprints are to be useful if there are 30+ fingerprints associated to it. Please note that I cannot 100% say this is because of the duplicate ISRC, but upon finding the issue of a large number of IDs, I also find the duplicate so there is def a correlation, which makes sense, since real remasters do in fact sound different.

This appears to be the edit making this:
I do not intend to place blame or anything at all, I am sure I have made more than my share of edits like this in the past. I am just noting the behaviour patterns on such things, this was also added separate (remaster vs non) and was combined after the fact.

It was decided that remasters would not warrant new recordings except if you hear a significant difference (making it close to a new mix) then you would also explain this difference in the recording disambiguation comment.

In other cases (almost always) we have all ISRC on same recording.


I have changed the title to something more representative of the topic.

If that is how the community wants it, that is fine. But then there is no need to worry about things like credits on recordings and acoustIDs. If a title and artist name and duration matches, it is a duplicate and should be merged. Also, the mastering relationships will be mixed or wrong. On a remaster of this type, the mastering credits will not be the same, making the relationships of such recordings wrong, unless maybe only the original version counts, not sure the logic in that so it will need explaining.

This will mean a reversal of a LOT of edits and adds. It was said prior that things like a DJ or party mix where track A is blended to track B causes a new recording. Hopefully we can get an answer here on this that can be referenced for mass merges.

Not at all - some can be live, or remixed, or dj-mixed.

Hence why mastering relationships are not allowed on recordings anymore (any left are just left over and not moved yet) :slight_smile: (which yeah, it’s data loss to some degree, but it makes shared relationships easy and was also decided because most of the time it’s not even possible to know which remaster is used in a compilation for example).

It does, because the sound changes significantly. Most remasters have insignificant sound changes and can be ignored. Otherwise, they’re not really a remaster anymore, they’re effectively a remix (and a different recording). See the relevant guideline.


Well, yes. I consider that part of the title, like “Song (remix)” vs “Song”. I am excluding different titles in that statement.

A fade in and fade out I do not consider significant at all as the actual song itself is no different, it is just the few seconds at start and end that differs. There are 2 different types of a remaster. What you said is correct as well, noting the 2 types, and that is my exact point here. When a remaster goes back to the raw recordings and redoes the release, that is significant. If it is just a move of medium, that is not. The difference being restoration and other criteria as outlined by ISRC and stated in either this forum or the edit. There are also remasters that are so different they are noted here: https://en.wikipedia.org/wiki/Loudness_war. Another editor has pointed this out as well.

For reference:
A new ISRC must be issued when […] A ‘full restoration’ of a historical recording is performed by re-mastering, re-pitching, re-equalizing, de-noising or de-clicking a sound recording to meet contemporary quality standards. The registrant of the recording has discretion when making the determination between full re-mastering (sound restoration) and simple re-mastering (reproduction without sound restoration).

New ISRCs are given for completely random reasons - I know what the rules say, we read them while deciding whether to allow multiple ISRCs per recording, but almost nobody follows them with any care. They’re just happy to call everything a full restoration.