The problem with some tags I use is that there either is no conventional tag for that information, or a conventional one comes close but the definition is subtly different.
In the former case I just have to make one up anyway. There’s info I wanna capture that has no existing convention to speak of.
In the latter case, “misusing” a tag would just cause further issues like the value being displayed in the wrong place or used for logic that goes wrong. Some tags like the movement ... ones kind of belong together and probably have conventional rules/expectations about their combinations and values given their intended use for classical music, or players will assume/enforce such because “it just makes sense”.
I’m considering some mass retagging currently, I already use some id3 fields anyway (title, album, artist, etc because their name matches in foobar’s mapping) so I might as well use as many as I can and reduce my exceptions. On the other hand, since foobar’s mapped names for id3 tags are unique to foobar and used directly for non-id3 tags, my non-id3 tags would get polluted with names that both go against their convention (abbreviated/concatenated words) and wouldn’t be read by other players either. So the more I respect id3, the more it breaks my flacs/oggs and binds them to 1 player.
In the end I chose “just name it naturally” because everything else, even id3, is just some level of vendor flavor.
This has been an interesting discussion though, thank you. My library setup is over a decade old so it may be time for innovation of some kind. Sorry if the rubberducking has been a bit much but it helps to write about a problem.
On that we agree. When I wrote a script to import and export lyrics from and to audio and external lyrics files recently, getting the SYLT id3 frame to work properly was infuriating and I would not recommend using it.
As long as you’re aware of it and can configure your tagging software to transparently map things correctly so you only have to deal with one human friendly field name per tag, a mix is probably the better solution when wide support is the goal.
That’s still better than what I did when I started collecting digitally 20ish years ago. I slapped most of the extra information I wanted in the album tag as that was supported by all players at the time.
I could easily extract it and sort it into correct fields these days but going through 10s of thousands of albums, even with filters and regex, is rough.
I personally try to respect the correct tags if they exist, otherwise I use custom vorbis tags for the same reasons as you.
If you want to perform changes at scale I can’t recommend mp3tag enough. It has powerful filtering, scripting and actions that can automate many things easily if you’re good with regex and basic programming/scripting principles (which I assume you are).
I’ll probably have to do that with my library at one point as well, but I’m still waiting for a time when it fits on an SSD in its entirety (that doesn’t bankrupt me) so I don’t go mad waiting for each step to complete on spinning rust.
I found this entire thread illuminating and interesting.
Take TRACKNUMBER from the foobar id3 mapping. It’ll be mapped to TRCK in id3 and trkn in mp4, but be TRACKNUMBER in vorbis/ape. But then mp3tag says TRCK/trkn map to TRACK in its user variables, meaning it probably reads TRACK from vorbis/ape. So vorbis/ape track numbers set by foobar don’t show up in mp3tag just because they have different user mapping names and there’s no standard for vorbis/ape to map to. Without testing this right now in a default foobar and default mp3tag, I’m pretty sure that’s the exact problem I saw years ago, tags showing up in one or the other but never both.
I could fix that with a custom mapping but it’s an endless problem with every player and tag editor I’ll ever use and I’m bound to forget a tag at some point. I went down that rabbit hole years ago and found it unmaintainable, so I just stay away from the inconsistently mapped standard fields now and only use tags that either coincidentally match all the time (like title, album) or custom tags that’ll always be mapped verbatim.
I only felt the RATING tag is ubiquitous enough that it would be worth asking to universalize its support, the rest is a me-problem that I avoid by having only vorbiscomment/ape in my files (and thus consistently (un)supported tag names) and using players with enough flexibility to read my fields.
But maybe I have it all wrong somewhere, or aligning the mappings once isn’t such a big deal now that my tags are stable I’m actually thinking of experimenting a little again now, setting some standard and custom tags in various ways and checking with a hex editor what really gets written and read from and to the files. I’m surprised for example that you wrote that an id3 TXXX:*** field name would get written to flac’s vorbiscomments.
I agree that the bungle of internal mappings and format specific tags can be both confusing and annoying. What I usually do is go the path of least resistance. Unless something breaks I try to use the default mappings. However my library is pretty much 98% flac and the rest mp3s with a few lonely mp4s mixed in so I luckily never really had to dive into the format specific mappings at scale.
Since it works with the software you use that’s a very neat way of handling things. When tagging, have you scripted mappings of existing tags to your own tags or how do you convert new additions to your scheme?
As long as you have one piece of information per tag, shifting around the tag names and mappings is trivial in mp3tag (if your library is not too big to be opened in one go). But since your system works with the players you use I don’t see a need unless you enjoy experimenting and want to “do it right”.
That was an error on my part. I haven’t looked at a flac file in a hex editor in a while. They’re simply saved as key=value pairs in the flac file as expected.
The tags of mp3 files with utf-16 id3 2.3 also waste a wee bit of space compared to vorbis tags.
I do it manually for every new file/album, can’t automate for the arbitrary state of files I take in. The foobar Properties dialog being able to bulk-copy/paste values between tags and tags between files, and showing both mapped and unmapped tags together, is a lifesaver. Just shuffle some values around, delete some noise, add some data from a quick wiki search. For string manipulation like stripping leading zeroes or splitting names on , / feat. & I use masstagger if it’s in bulk. I really only use mp3tag if I want a look at my files without the bias of my customizations.
Yeah, and the editor-dependent naming of the mapping is the problem there. I used to bulk convert and retag a large part of my collection to my external players like a pmp or my car’s usb stick, but it’s a concept and chore I never liked, including discovering exactly what each device expects in tags and things like cover image naming/format.
Lately I’ve instead been syncing my music to a share on my nas that my phone can get to by vpn, and searching a good android player with android auto support. Mobile data subscriptions in this backwater country () are finally catching up with the world in terms of price per GB, so streaming from my server is an option now. The nas comes with its own player, but it’s junk in terms of ux/ui and I really didn’t like it only supporting cover.jpg (and not .png). Whichever player I choose will have to suffice for years because I’ll do the bulk tag changes at the source now and just adapt my foobar for them, which I really don’t want to do often.
Tried symfonium’s car player yesterday and it was
Symfonium’s UI just makes a ton of sense, the way it aggregates and structures info. Loved seeing things like the identical date on all tracks being hoisted to the album info, or an artist leading to all their albums and the individual tracks, with cover images just working.
I’ve ironed some exotic formats out of my collection already and will bring my tags more in line with standards when the next release is out (at which point it should all work, thanks again @Tolriq), then finish the switch and donate.
The only thing that won’t work is my midi files with ape tag sibling files, but oh well, that’s so niche I don’t count on it working anywhere else.
I’ve created actions in mp3tag for all of these cases (I’ve got around 30 in total I think, sorted into submenues by which field they affect). The filter options are also really powerful if you only want to edit a portion of the files.
I usually first inspect all the tags the files have (or rather all supported by mp3tag to be precise) with the extended tag panel in mp3tag (alt+t) and delete tags that I do not use.
Then I use a websource plugin to look up the metadata in musicbrainz. If it’s not there, I see if it’s on discogs instead and use the built-in websource to get correct metadata that way.
Then I adjust the resulting new tags with the custom actions to my personal conventions and lastly use an action that creates new folders and file names based on the new tags while also copying over the other files in the album folders (lyrics, logs, cue sheets etc.).
Once that’s done I run my script flacr on the new folders which recompresses all flac files, sets a fixed padding of 4kb and calculates replaygain tags for all music files (not only flac). Once that succeeds without errors the files can be copied to my nas for MusicBee, Jellyfin, Plex, Mediamonkey etc. to be picked up on the next scan.
For that I use MusicBee and set appropriate synchronisation rules per device. Embedding covers, lyrics etc., converting on the fly or even using a fake “virtual device” if the device cannot directly be synced to (like a car player) work really well and I don’t have to change anything in my music files. The only device that approach doesn’t work for is my ancient iPod nano from 2007ish. For that I first export a playlist to a virtual device (a local folder) in MusicBee and then use foobar with a plugin to sync the resulting playlist to the iPod. I could use iTunes but I dislike it too much to install it.
Tell me about it. In Germany they charge outrageous prices for mobile data. For example 15€ per month for 5GB data. My home connection can download 5GB in 75ish seconds. It’s a joke, just not a good one. I usually use Symfonium to offline cache a few albums before heading out so I don’t have to rely on mobile data at all.
I tried a few different server solutions as sources for symfonium. Jellyfin and lms have worked best for me in terms of supported tags and performance, lms being more lightweight and purpose specific than jellyfin. If your NAS can run docker containers, I’d suggest you give lms a try. Navidrome also worked fairly well but the multi value tag support broke its neck for my personal use case.
What convinced me was the performance with my library size. Jellyfin or plex love to think 10ish seconds before reacting to a query. Symfonium still feels fairly snappy thanks to the offline db. It was the first and only android app where using my library was enjoyable instead of painful. The only real pain point that remains are the sync times (takes 50ish min per sync for me) but hopefully eventually one of the servers will implement a delta sync to resolve that as well. One can only hope and wait.
Funny, I didn’t know that was a thing. To be honest the only times I’ve come into contact with ape tags so far was when people in the mp3tag forum were complaining why their tags did not work and the answer was usually that there were ape tags as well as id3 tags in their mp3s, leading to conflicting results in some players and the solution was often to delete the ape tags.
I’m actually thinking of experimenting a little again now, setting some standard and custom tags in various ways and checking with a hex editor what really gets written and read from and to the files. I’m surprised for example that you wrote that an id3 
) are finally catching up with the world in terms of price per GB, so streaming from my server is an option now. The nas comes with its own player, but it’s junk in terms of ux/ui and I really didn’t like it only supporting 
