Advanced Auto Playlist assistance

[Phaedrus]

My specific goal is to create an auto playlist that will show me albums that I have more than one copy of. For example, I buy a CD I forgot I already had, re-rip it, tag it (tags are slightly different because it's a different version) and import it. MusicBee will happily import it, and copy it to the location, but not find that it is a duplicate because the titles are slightly different.

In order to do this, I would like to create an Auto Playlist that contains albums which have two "Disc-Track# 1-01," and I can't figure out how to do this. If it's even possible. I'm approaching this like a SQL query, and that might be my problem. What I really want is to use the "Having" part of a query.

Any thoughts on the matter will only help my auto playlists get better and I really appreciate that.

[redwing]

For that purpose, I'd rather use column browser than auto-playlist.
First I'd create a virtual tag "<Album Track Count> - <Album Duration>" and then add the field to column browser as first column. Second column is <Album> and third column is <Album Artist>.
Then sort first column by track count, and check items one by one from the top. If an item has multiple albums, second and third column will tell you.

[Phaedrus]

Thanks, I really appreciate the suggestion. I spent some time with this because I was so intrigued. In the end, I don't see how this helps. It seems like it only catches identical imports, and misses a lot of potential scenarios;
1. Same album with an extra second on one of the tracks wouldn't register in any of the other columns
2. Same album with "part one/part two" of a track split differently wouldn't register as the same in the other columns.


My brute force method was to use the standard three columns, and sort the tracklist by "Disc-Track#." This let me scroll through the albums quickly, and watch for a second "1-01." The display takes less than a second to update, and all I had to do was stop hitting the down arrow when I found one, and analyze the entries.

This method seems to take a bit longer to update the columns and on the multiple albums with identical times and total tracks, (I was surprised at how many there are.) I need to spend extra seconds to determine if I need to go further. With 6,752 albums, every second counts.

Assuming 1 second per entry.... either of these methods would take hours to complete.

[hiccup]

Another approach might be to batch run all your albums through MusicBrainz's Picard.
All your albums (if present in musicBrainz's database of course) will get elaborately tagged, including a release-id.
That release id will make it easy as pie to find all duplicate albums.

But if you have no interests in MusicBrainz data other then for this specific purpose, it would be like shooting at a mouse with an elephant gun.
Else you might be catching two bees with one stone...

[redwing]

In the end, I don't see how this helps. It seems like it only catches identical imports, and misses a lot of potential scenarios;
1. Same album with an extra second on one of the tracks wouldn't register in any of the other columns
2. Same album with "part one/part two" of a track split differently wouldn't register as the same in the other columns.

Of course you should adjust the suggestion to your particular setup and purpose. I am also doing the same thing with my collection and it helps me. That virtual tag is independent of any tag values working like an unique album ID, which I'm finding useful. Also you can find albums with 1-2 seconds difference in album duration just above and below the item if you sort by the field.

[Phaedrus]

Another approach might be to batch run all your albums through MusicBrainz's Picard. All your albums (if present in musicBrainz's database of course) will get elaborately tagged, including a release-id.

This is a good idea, and has given me a couple of new directions to think about.

Sadly, part of my problem is my fetish for tags. I run all my albums through Media Monkey, and the Discogs plugin. I confess I haven't looked at MusicBrainz in the past couple of years, but I recall having many albums not found, and less information when it was in the database. With Discogs, I end up with all kinds of neat, (although admittedly ultimately useless) information. I do get an ISRC number though.

From the fuzzy corners of my memory, I had a couple of albums which were distinct releases, with the same MusicBrainz ID.

[Phaedrus]

I like the idea of combining both of these tag ideas to form a "super album ID." If I combine track, total time and ISRC into one Meta-meta-tag, I can come close enough to a guarantee that I should only have one result.

This would find the double-imported albums. Now, I just need to find the ones that are just slightly different.

[hiccup]

Another approach might be to batch run all your albums through MusicBrainz's Picard. All your albums (if present in musicBrainz's database of course) will get elaborately tagged, including a release-id.

This is a good idea, and has given me a couple of new directions to think about.

From the fuzzy corners of my memory, I had a couple of albums which were distinct releases, with the same MusicBrainz ID.

The boys and girls at MusicBrainz are very ocd about their database, and the structure is very well thought-through.
One aspect that is clever, and (for as far as I know) Discogs doesn't do is this:

Album releases from different countries, remasters etc. will get different album id's, but they will all get the same releasegroup id.

The same for songs:
Studio and live versions of a song will get different track id's, but the same work id.
That even goes for the same song recorded by a different artist.
So that makes it for example also very easy to find all recordings of a song, irrespective if they are live, studio, radio or album releases, and also cover versions by other artists.


But MusicBrainz's database is populated and maintained by volunteers. So the completeness will depend on their effort and musical interests.
Discogs has it's origins in some alternative commercial market place, and seems a lot less concerned with having a well-structured and dependable database.

It might be worth to give Picard a renewed try.

[frankz]

The Picard idea is a pretty good one, actually, because you don't need to wind up writing its tags to your files ultimately after the scan.  

You could have it scan the files to identify them and then it organizes them into its tagging pane.  Then you can visually scan that pane for albums with a result of - for example  - "(24/12)" (meaning that album has 24 files identified and 12 tracks).

Of course, it's not perfect, but even if it doesn't find a number of the albums in its database, it'll knock down the number of albums you'll have to locate using other means.

[frankz]

The boys and girls at MusicBrainz are very ocd about their database, and the structure is very well thought-through.
One aspect that is clever, and (for as far as I know) Discogs doesn't do is this:

Album releases from different countries, remasters etc. will get different album id's, but they will all get the same releasegroup id.

The same for songs:
Studio and live versions of a song will get different track id's, but the same work id.
That even goes for the same song recorded by a different artist.
So that makes it for example also very easy to find all recordings of a song, irrespective if they are live, studio, radio or album releases, and also cover versions by other artists.


But MusicBrainz's database is populated and maintained by volunteers. So the completeness will depend on their effort and musical interests.
Discogs has it's origins in some alternative commercial market place, and seems a lot less concerned with having a well-structured and dependable database.

It might be worth to give Picard a renewed try.

Picard is fantastic at what it does and Discogs is fantastic at what it does.  Their intents are different.

You've identified MusicBrainz' strengths well - they cross reference songs and releases and are very focused on those things.

Discogs is a different beast.  Discogs aims to identify and document each pressing of each release.  They do have a unifying identifier for a release like MBrainz, but they go deeper into detail as far as this specific record was cut in Jacksonville VS the West Coast and pressed at this plant, etc.  Etched matrix numbers and run-out grooves and that kind of thing. 

This is why I'm slightly confounded when people use Discogs as their go-to for file tagging.  Their data is good in that respect, but the meat of their data and their focus has no bearing on tagging whatsoever.  Of course, they do have the data for it for the most part, but it's not really what it's built for.

I can verify that their database is also maintained by volunteers and that they are, indeed, quite adamant about data quality.  Quite.  Obsessively so.  Conversely, I've seen some real garbage data in MBrainz that has gone unaddressed for a very long time (certainly not the norm, though).

That said, I prefer MBrainz for our purposes here.  Their data is much better suited for file tagging, and, because they have an house program that does all the work for the end user, you're not relying on third party plugins like you do with Discogs.

[hiccup]

Indeed both have specific strengths and weaknesses.
One thing that made me focus more on MusicBrainz than on Discogs is that if you are e.g. interested in artists performing on an album, you would have to go through all releases, because the album credits would differ completely between the different releases (cd/lp/regional etc.)
There doesn't seem to be any syncing of such information put in place?

Also, there are scripts and plugins for Musicbrainz/Picard that make it easier to retrieve info from Discogs, so that would also help getting best of both worlds.

@Phaedrus:
Drifting a bit off-topic here. Just say stop if it starts bothering you.

[Phaedrus]

Suddenly my beard is itchy; I'm scratching my chin pondering these things.

It was several years ago that I settled on Discogs for my tags. I knew at the time that it was imperfect, but it was light-years ahead of MusicBrainz for tagging information. I honestly think it was the plugin available for Media Monkey that made the difference. Once I made the decision, inertia took over, and I stopped looking at any other tagging solution. Now, I have two people with carefully thought out and presented arguments for the same thing. This definitely makes me think twice about my current process.

After re-visiting the MusicBrainz site, I'm reminded of one of the items which kept me out. I don't see where MusicBrainz includes "Involved People," which is one of the fields I really wanted. As I said before, It's neat, and ultimately useless, but here we are. I'd like to say that it would be a no-brainer to just stop getting the information, but it just doesn't feel right to leave data behind. Maybe I need a paradigm shift.

*Edit to add; I'll have to play with the MusicBrainz plugins to get everything together.

In the end, it seems like it's worth it to run one of my old backups from my recent upgrade to 4TB drives through MusicBrainz to see what I get.

@Phaedrus: Drifting a bit off-topic here. Just say stop if it starts bothering you.

Please, carry on. I always enjoy a good discussion. I really would like some help with the playlist at some point, though. I suspect that if I can get this bit sorted out, it would answer many of my other auto-playlist wants. I simply refuse to sort through my entire library Album by Album ever again. I know the relationships I want to make between tracks / Albums; I just don't know how (if?) I can make them happen with an Auto-Playlist.

[hiccup]

After re-visiting the MusicBrainz site, I'm reminded of one of the items which kept me out. I don't see where MusicBrainz includes "Involved People," which is one of the fields I really wanted.

You mean like the people/roles such as mentioned here? :

https://musicbrainz.org/release/e9fd5395-94a3-4c0c-8c7e-edb8e131cdfa

I really have no idea if Discogs does a better job on this, or is more complete in a general sense.
This specific album is of course a very popular example that is prone to be very complete in this regard.

Why not just pick 10 albums that represent importance to you, copy them to another drive, and run them through Picard?
I am certainly not advocating MusicBrainz is some golden standard, and I am also interested and curious about your findings, which also might be interesting to others following and considering all this.

Just for my information, how are you retrieving tags from Discogs?
Through MusicBee or by means of a 3d party application?

[frankz]

I would urge you to download and install Picard and try it on a few albums rather than picking through the MusicBrainz website for data.  Like I said, you don't ultimately have to tag the files with the data it retrieves, but you can see your old data side by side with its proposed changes and compare the two for anything missing.

I don't think they call them "Involved People" per se, but Brainz does have a list of artists, musicians, producers, arrangers etc for the tracks in its database (the ones that are properly tagged).

Specific to moving forward with your original question, you could tag your files with just the proprietary Brainz tags (their album id or release id or album artist id or some combination) that will help you identify duplicates you have while leaving the Discogs retrieved data in tact. 

If your stuff is already tagged I wouldn't recommend retagging it with MBrainz.  The data isn't that radically different or earth-shatteringly better.  But going forward, especially in the interest of abandoning MediaMonkey and its plug-in if that's your intent, it's something to look at.

[hiccup]

To complement frankz good info and tips:

Just for the purpose of your original issue, you could setup Picard to only write the releasegroup tags.
Just be warned that Picards learning curve is a bit steep (or long, or erratic, what should you call it) to accomplish such refinement, and you will need some scripting for it.