Year is wrong

Hi Noticed that quite a few entries in a series have the wrong year.

Here is an example.

The year in the meta data of the release is 1988, as it an edit for that song, released in 1988.

However someone has submitted is as 2018, which is the release of that album. This is in-correct according to the publisher.

However when I try and edit it, I can’t seem to change the year. The genre is also incorrect.

But I can’t seem to change that either.

It seems most of that series is incorrect. as per the details on the publishers files.

It’s complicated.

Firstly recordings are a highly duplicated dataset. If you look at the list of the recordings of “The Look” you can see that there are about 70 records for recordings of this work by Roxette, many of which have duplicate recording lengths and even remix versions with almost the same recording length which are almost certainly the same actual recording and remix, just slightly different lengths due to fade in/out or track gaps.

Secondly the date shown is the date of the release the recording appears on - and is not a Recording date as such…

If you edit the Recording and then edit the Work relationship within the recording you can enter a start/end year of 1988 as follows:

I have done it for this particular recording so that you can see what the update looks like.

I would have to look at the code to see whether Picard tries to work out what the recording date could be (e.g. perhaps as the earliest date of the releases that the recording appears on). What metadata variable are you using for this?


I checked the code. Picard gets the recording first release date from Musicbrainz, and the ticket which explains this is here.

The date is calculated as the earliest date of the releases it is a track on, and it is the recording dataset pollution which causes this to be wrong.

The way to correct this is to have detailed knowledge to make safe and accurate merges of recordings where you know for certain that the two separate recordings on Musicbrainz are actually duplicates of the same actual recording/mix i.e. if you listen to the two tracks, they would sound identical in every respect. Length within a couple of seconds is a good indication so, for example, judging by the length alone would seem to be a duplicate of

Cleaning up the recordings dataset for the literally billions of recordings therein would be a seriously non-trivial effort.

Sorry to be obtuse but I think you’ve used this attribubte incorrectly to appease the original request.

Firstly, with this specific recording it is part of a DJ mix which means it will almost always create its own unique recording (even if its a song released previously, it’s likely this recordings general structure has not previously existed or will exist ever again outside of this particular mix).

Secondly, the date you’ve put there (the work date) should be the date in which that specific recording was recorded/produced etc. This is almost always an unknown unless there is some kind of written evidence to prove. For example, many Jazz releases will state a date for which that recording was recorded - that becomes the work date (for example Song “Gravy Waltz” - MusicBrainz). Live recordings also usually give a partial or full date to work from, this becomes the recording > work date. Many “pop” (for lack of a better term) recordings do not bother to keep this - this can because they don’t feel it is important information for the listener, or because the track actually took multiple days, months, years to finally become so they don’t actually know when the recording date exists.

With this example it looks like the DJ Beats “remix”, only exists as part of this DJ mix. As this DJ mix was released in 2018, that recording will be forever seen as being released in 2018. You could on a limb guess that the entire DJ mix was recorded sometime in 2018, but unless you have the paperwork to back it up it’s nothing more than a guess. At no point did this DJ mix exist in 1988, unless again there is evidence to the contray. Looking at it, DJ Beats doesn’t really seem to exist anyway and looks to be a generic alias used by “MasterMix” - and with a name as generic as “DJ Beats” I can’t find anything on the internet that links this to an actual person.

Sadly OP can’t really get around this unless they wanna try some trickery with Picard scripts which could go out and look for the earliest version to exist via the work relationship. A script already exists to find the earliest recording date for a recording, but as previously mentioned the earliest version of this specific recording is 2018.

I’ll be reverting that change as its ultimately incorrect. For all intents and purposes that recording is from 2018, not 1988.


Actually on this, this series is a complete mess that I’m unfolding.

You’ve got releases marked as DJ mix yet they’re not a continuous mix of music (they’re just intended for DJ use but otherwise unmixed). No confirmation that these ever existed as a physical format except for a few “trust me bro”'s.

Along with that DJ Beats isn’t anyone, if anyone it’s going to be in-house guys like Jon Hitchen or Richard Lee.

However it still stands, this recording has been amended from the original. When it was amended is hard to say but it cannot be assumed it was 1988 or 2018 (as there is no proof of either); so Picard will just default back to the release date (2018)

1 Like

Except if you look at the meta data which is released on the original, as it’s a digital release it will show the year as 1988

Here is a screen shot of the latest release showing the dates on the files.

In additional, almost all the genres are incorrect.

So if someone changes 1% of something it’s a new release?

But you’re correct for the other things, DJ Beats is a series, that’s released to help DJs.

they all started as physical, but now are solely digital.

Maybe, but that’s the metadata from Music Factory (MF). Just because MF say “1988” doesn’t mean that it follows our guidelines and vice-versa. I can take anything and put whatever I decide in the year field, that doesn’t make it a factual statement. We have this issue with Spotify, who like to indicate that digital releases existed en-masse before ~2000 which is obviously incorrect, and so you’ll find Musicbrainz contributors doing the work to find out the real release date.

MF is likely providing the original date that the recording was released as a UK single, which is great for them and for those who need that information but doesn’t align with the way MusicBrainz stores and presents the data. The fact is that recording almost likely was produced/recorded/mixed/finished etc. in 2018, thus in MusicBrainz world it is a 2018 recording, even though it is based upon a 1988 recording.

As for genre’s… genre’s are not factual in the eyes of MusicBrainz, they’re considered subjective. Some people like to get into the weeds and tag things down to the latest unique strain, and other’s are happy to have things tagged as pop or rock. The main thing is that this allows people to chose the genres they want to use in their libraries. For that reason sometimes genres may appear wrong, because the genre you expect for a piece of music may not align with someone elses.

In Picard under Options you can set the behaviour around genres. See mine for example:

So with this I’ve told Picard to only use my genres, don’t fall back, allow Folksonomy (which means tags that aren’t classified as genres in MusicBrainz) and to restrict some Folksonomy tags in the box below. I’ve also stated it to not include anymore than 5 possible genres.

You can adjust these settings on your own Picard installation to your tastes. You can also submit your genres back to MusicBrainz from your local files to try and help others with identifying their music,but and that might be where the wrong genre is coming from.

The quickest option might be to simply down-vote the genre on the recording and that should strip it from the recording view in the web browser, and thus update in Picard (but someone might be able to correct me on that one).


Pretty much yeah, it’s a bit of a grey area.

For these certainly these are new recordings (as they’ve considerably adjusted the structure of the song by adding an intro and outro, thus influencing the duration of the recording).

New recordings are usually avoided if say the duration is different due to silence, or if it’s a particularly bad transfer, for remasters (so long as they don’t adjust the structure of the song) or for any variances in track “volume” (which is usually caused by mastering).

However that is always up for discussion and debate as to what constitutes a new recording or not.

1 Like

Yep looks like an evolution of DMC (DMC - MusicBrainz) - these labels are certainly cool to document, but the thing is to many they’re really obscure because of their limited reach - another example would be BBC Radioplay Music service (BBC Radioplay Music - MusicBrainz)

Indeed, and unfortunately I can’t find at what point they cut all CD’s being issued as it would be useful to know; maybe something for another evening :slight_smile:

1 Like

It’s interesting as actually they’ve probably changed less than 10-20%.

It’s effectively a safe intro and outro area for DJ’s to mix into and out of easier. Most of them are a similar running time to the main songs.

So if I wanted to use picard to get the earliest recording time, as per the meta data, what’s the best way for me to do that, to reflect the data that’s coming from MF?

As that’s the information that’s mostly useful to a DJ.

I mean if you’re playing a 00’s gig, people are going to look at you very oddly if you play Chuck Berry - Johnny B Goode, just because it was a CD released by someone somewhere in 2009…

As to the switch from actual CD’s to purely digital, I can get that for you.

oh no doubt about that I understand why MF produce these, and their purpose

Sadly I’m not the man for the job, maybe someone else might be able to do some script-voodoo that can find the date information you desire

Yep I understand, however keep in mind that although MusicBrainz database can be useful to a DJ, its primary purpose is not for a DJ’s requirements. It exists to be a record of music entities (recordings, releases, artists, events, labels etc.) with metadata surrounding it in a structured and organized format that links it all together. Pretty much it’s sort of like a library.

Indeed which is why I’ve seen some DJ’s organize their collections meticuously by their own standards, not using the “best practice” of a community of people. It’s like if we were going back to physical records, and I came along and went “no no no, you should order your records like this” it’s likely you’d probably kill me because I’ve just gone and put a particular record in the wrong box and you don’t know where that is, but to me it makes sense.

Yep if you’ve got an article or a press release or anything like that, would be another thing that would be useful to note down in the annotation field :slight_smile:

1 Like

Yeah I guess this is more like you can order your set of music how you’d like and I’d like to do it too.

but to go back to your library analogy, even books have a first published date in them. But I guess it’s hard when as a DJ I’m more interested in the “single” than the album.

Is there a way to get MB to treat it like a single or return a single instead of the album?

Also if you are going to revert back, could you leave the original recording date as 1988, as per


Unfortunately not, that field mentioned in that JIRA ticket is automatically calculated, no one can overwrite it as it will just use the earliest date of the available releases.

Ah that’s a shame, as it doesn’t seem to be working it’s way down to the client.

hey ho, thanks for looking.

This is NOT the case. I did not add this attribute incorrectly “to appease” anyone - I added it because at the time I thought that it was correct. Since people cannot know what my motivations were without asking me, I would very much appreciate it if people didn’t accuse me of motivations which I never had.

That said, I agree with…

and I would have reverted this change if he had not already done so.

The above statement accords with my own thinking. These are unique recordings created with a specific intent to be used as continuous music, and so do qualify as DJ-mixes. My detailed analysis of whether these releases are DJ-mixes or not, together with the supporting evidence, can be found at Edit #104490387 - MusicBrainz.

However, despite my analysis which has not responded to, and despite the above statement, Dan has instead decided that these are NOT DJ-mixes after all and has gone ahead and made automatically applied edits to remove the DJ-mix tag from ALL the series releases without allowing the community to vote on them.

Hi, please understand that I am not chained to my PC constantly, and have only logged in this morning to review your written comments from the notification I recieved from logging in.

Let’s break this down.

Firstly, you can easily find the pages for these releases are easily calculated - example DJ Beats 21 (Remastered) - Mastermix just change the number at the end and you’ll bring up the page for the ID in question. Apologies for not including those in my notes; and once we’ve worked out the released format issue with this series I’ll add the URL’s for each entry in the database with the correct relationship type.

Secondly, no I still do not believe your understanding of the term DJ mix is correct. If you look through the database you’ll see that DJ-Mix is nearly always used for continuous mixes of music; Search results - MusicBrainz - a series of unmixed tracks, regardless for their intended purpose, is just a compilation and by no merit a DJ mix.

Thirdly, yes that page is correct if you just scroll down a bit you’d find that DJ mix is listed on that page as per:

A sequence of several songs played one after the other, each one modified so that they blend together into a continuous flow of music. Common techniques for this include crossfading, beat-matching, beat juggling or scratching. A DJ mix requires that the tracks be modified in some manner. To describe a ‘DJ mix’, use the DJ mix relationships. Note: If the tracks have not been modified, then the action which should be described by relationships is compilation, not DJ mixing. Also see medley, for cases where the sequence of songs involved a brand new performance, rather than the use of prerecorded material.

The page you have linked states pretty much the same thing;

A DJ-mix is a sequence of several recordings played one after the other, each one modified so that they blend together into a continuous flow of music. A DJ mix release requires that the recordings be modified in some manner, and the DJ who does this modification is usually (although not always) credited in a fairly prominent way.

None of these state “because the release label says its for DJ’s it must be a DJ mix” - no it’s a description of what other may consider a continuous mix, and is to catalogue relases that are genuinely mixed together for seemless listening pleasure such as those produced by Ministry of Sound, not “adjusted” so they can work nicely in a mix.

Per your note here Edit #104490387 - MusicBrainz you’ve just re-affirmed that these tracks are simply intended for DJ use, not that they are a DJ mix. At no point does that say that the individual tracks are mixed up together to make a continuous stream of music.

Your last note is unfounded too because you’ve tried to provide evidence for something existing that doesn’t exist:
That Amazon Search link returns a series of releases that don’t even correlate to what we are talking about:
For example the top three:

The first one is a series of megamixes (typical Music Factory fare) not the additional 8-bar intro/outro of the series we are discussing here
The second one doesn’t even appear to be from Music Factory, but instead Universal Music
The third one is much like the first and is again a series of megamixes.

Even as I flip through the pages, none of the releases given here actually prove anything because they’re not for the releases we are currently discussing.

As previously mentioned in this thread, Ian has confirmed that CD’s did exist but they stopped at some point but I need to find where / if Music Factory ever made this a public announcement then we can determine of these relases which are likely genuine CD’s or have been mistakenly input as CD’s yet only ever existed digitally.

We shall wait to hear from others what they think DJ mix means, and if necessary I will log a ticket to update the documentation to make it clearer; but for the moment I will be making no more changes in regards to the secondary format of these entities before we end up in a tennis match of edits.

1 Like

On the format front, so there’s a seller on eBay UK who seems to have multiple copies, and they’re photographs of a physical item:

Specific example:

yeah, in my experience, DJ-mix has always meant “continuous mix”, not a mix for DJs. I would be for clarifying that in the documentation and even a new secondary type for these “mix for DJs” type of release

I also don’t think there’s currently a way to do that, but it might be possible to write a Picard plugin for that, if you have it follow the Edit relationship back to an older recording. I wouldn’t do it through works, as there’s many ways that could go wrong, like getting the original (non cover) recording date for a later cover

1 Like