Author Topic: The Road to Perdition, or how I Stopped Worrying and Learned to Love MusicBrainz  (Read 16919 times)

Phaedrus

  • Jr. Member
  • **
  • Posts: 117
The general sentiment among MusicBrainz veterans is "any data is better than no data" but "no data is better than bad data".

I respect that. Database Boyz, Represent!   ;D
Dell Inspiron, i5, 16GB, 256GBSSD
Windows 11 Preview
MusicBee 3.5.8447, standard install on C:\
Library on External drive;
293,621 Tracks, 18,393 Albums, 4461.93GB, 850d 12:20

Topping D10 32bit 384 kHz USB DAC
O2 Headphone Amplifier
beyerdynamic DT770M 80 ohm
Crossfade M-100 Master Over-Ear

hiccup

  • Sr. Member
  • ****
  • Posts: 7790
Add them!  There's even an "Add Cluster As Release" plugin that makes it dead easy.

And if you become a MusicBrainz editor, you could use user scripts for your browser to import Discogs releases.
https://wiki.musicbrainz.org/Guides/Userscripts

F.y.i. almost 20% of my album releases are at this moment not present in the MusicBrainz database.

Phaedrus

  • Jr. Member
  • **
  • Posts: 117
Update on my process.

I'm working on the "Js" in my list of studio albums, and the process is so much better than when I was running them through the Discogs plugin in Media Monkey. I regret all the time I wasted doing that. I'll go a step further, and say that I really regret all the time I wasted using Media Monkey as a tag manager. To be fair, my tags are more consistent than I realized, at least in the Studio album section of my library. I can credit some of that to Discogs.

So, here's what I'm doing;
1. I'm selecting artists in the file structure in the left pane based on how many tracks I have by that artist as shown in MusicBee. Artists that have well over a hundred are processed separately.
2. Keeping the total tracks processed at a time to under a thousand, I drag them to the "Unmatched Files" section, and let them process.
3. When the albums are all loaded, I select all in the left pane, and drag them back to the "Unmatched Files" section, and "Cluster."
4. With the files all clustered, I do a "Lookup."
5. Go through all the results, and delete the albums that have zero tracks listed, and make sure that all of them are "Gold."
6. "Expand all" and scroll through the entire list and check out all the tracks that are dark red.
7. "Collapse all" and scroll through each label to make sure I like the cover art.
8. "Select all" and add them to my processed collection, and then save.

As I go, I'm finding some troublesome albums that don't seem to fit. I'll make a quick effort to fix it. If I can't, I create a shortcut to the folder in a "difficult" folder. Once I'm done with the first pass, I'm going back to see what I need to do to make them work. This is where I will add them to the MusicBrainz database.

I did have to keep MusicBrainz from modifying the Album Artist. I have several Album Artists that I change to be more generic, for example; Frankie Valli. I really don't want to care if he had the Four Seasons with him or not when I'm looking for a track he did. I was going to try to code that in, but I already had it in place in the Album Artist tags in my files, so why would I needlessly over-complicate it?

In all, I'm happy with the results of my little experiment. I've only played a little bit with auto-playlists on my processed tracks, and the results seem better. Not much difference in my browsing, but that just tells me I had done a pretty good job before. Nothing has jumped out as being outright wrongly titled, so I've got that going for me. Which is nice. I would definitely encourage anyone on the fence to just jump in. The learning curve is there, but it's nowhere nearly as bad as I had thought. Next time through will be way faster now that I have acoustids, and MusicBrainz info embedded.

So thanks for listening, and thanks for the encouragement. I wouldn't have done it without you.
Rock On!
\m/  ;D  \m/
Dell Inspiron, i5, 16GB, 256GBSSD
Windows 11 Preview
MusicBee 3.5.8447, standard install on C:\
Library on External drive;
293,621 Tracks, 18,393 Albums, 4461.93GB, 850d 12:20

Topping D10 32bit 384 kHz USB DAC
O2 Headphone Amplifier
beyerdynamic DT770M 80 ohm
Crossfade M-100 Master Over-Ear

hiccup

  • Sr. Member
  • ****
  • Posts: 7790
Instead of creating links to troublesome folders, you could setup one or more custom tags to populate MusicBrainz ID's for album, track etc.
That will make it easy to filter which albums/tracks still need work in that respect.
And then you could also use these tags to create custom links in the details tab to directly open the release page for the playing/selected track/album.

Another tag frame I wished I prevented Picard to (over) write earlier, is 'original year'.
I care for that one being the original release date for singles, or the first date an album was released to the public. (irrespective of it being a record or a cd)
But that doesn't always seem to be the aim of MuBr for 'original year'.
When it finds a compilation album for a single, it will probably write the first known date for that compilation album.
Or when it matches an original album to it being present in full in some box-set (which even might be correct), I believe it always writes the first known date for the box-set, not the date of when that album was first released to the public.

So while it will often will give good results, there are also situations where it might overwrite your carefully maintained 'original' years.

From that follows a piece of advice I would give other users starting to use Picard.
Begin with excluding most tags to be written (perhaps even start with only writing the MuBr ID tags), and then (slowly) learn which ones you need and then allow those.
That's probably better, and could save you from some regrets compared with starting the other way.

Phaedrus

  • Jr. Member
  • **
  • Posts: 117
I have my compilations separated from my studio albums in my folder structure. My first level of folders under my Music folder looks like this;
- C (Classical music; have to maximize the number of characters for files.)
- Christmas (Any Christmas music album)
- Compilation (Includes Various artists, and greatest hits albums)
- Live (Anything album that is all live recordings)
- Separates (Any single track that I did not get as part of a full album, or doesn't fit anywhere else. Kind of my "Untamed section," that gets sorted through half-heartedly now and then to remove tracks I've since got the whole album for. Everything here has an "Album Artist" of Separates to keep it separated from the rest of the library. This one will suck to "Picard.")
- Singles (Anything labelled as a Single in Wikipedia)
- Soundtracks (Currently including single artist soundtracks and scores, but I'm rethinking that.
- Studio (Albums recorded in studio as a single work, with above exceptions. This also includes EPs) The full path to music is; Studio\<Album Artist>\<Album>\<Track#> - <Artist> - <Track>

(I use a custom tag in MusicBee for <Album Type> that I set when I add the music from the Inbox. This corresponds to the first level so I can automate file organization.

I'm doing all my retagging in Picard outside of MusicBee. This is the reason I designed my folder structure the way I did. I can work with one type of album at a time. I also have the advantage of finding music easily outside of a Media Manager. For my Studio albums; I only deal with full albums. If a single track is off, I'll come back to it later. (I now keep my library entirely software agnostic. I use my Media Manager for playing, and other software for tagging. And never the two should meet. Media Monkey taught me that one.) It's way easier for me to work with Explorer shortcuts to get to single albums that may need to be entered in the database. So far there is only a handful.

I hadn't really considered the ramifications of release date until you mentioned it. I had to go through some gymnastics to get the "original year" to be the release year, because I don't care when the media was made; I care when the work was released. If I'm honest though, I never used the year, only the decade. It doesn't seem like much of an issue for the studio albums I'm processing now, but I will certainly need to deal with it for compilations. Lucky for me, I have a while before I'll get there. Glad I can avoid your trouble with compilations and Box sets.

Truly sound advice to keep everything, and add more later. Once you have the basics, it's really easy to go back through. The only reason I'm not following that is that Media Monkey really monkeyed with my tags, and they are a mishmash of versions, and the "Involved People" tag was pretty much hosed. The patch Steven did for Involved people worked great on almost all of my tags, but the tags where the Monkey wrote really weird stuff I have no hope of fixing. I liked the idea of stripping everything out, and starting fresh with good frames. I may have lost a few good bits of information in some of my tracks, but for the most part, I haven't lost anything I really use. (I'm also taking the opportunity to strip out the embedded lyrics. All of my lyrics will be separate files from now on.) Really, my only regret is not using Picard when I found it a few years ago.

I've found that this first pass is only taking as long as it has is because I'm getting album art. My second pass on Picard Processed tracks is way quicker, and I can do way more than several hundred tracks at a time.

I appreciate your insights. And as always, Thank you for your support.
Dell Inspiron, i5, 16GB, 256GBSSD
Windows 11 Preview
MusicBee 3.5.8447, standard install on C:\
Library on External drive;
293,621 Tracks, 18,393 Albums, 4461.93GB, 850d 12:20

Topping D10 32bit 384 kHz USB DAC
O2 Headphone Amplifier
beyerdynamic DT770M 80 ohm
Crossfade M-100 Master Over-Ear