Plasma Active Three released

Posted Oct 16, 2012 21:42 UTC (Tue) by sorpigal (guest, #36106)
In reply to: Plasma Active Three released by rsidd
Parent article: Plasma Active Three released

In despair over adequate tag handling I just gave up, years ago. Different players sometimes screw with id3 tags, (e.g. re-tag things like track, year, genre, etc) or use id3 in ogg, or worse. What's more, field lengths can be an issue, as can field counts, etc.. I tried folder hierarchies, too, but gave up on that: tagging is really necessary for cross referencing purposes. My current solution, which is highly sub optimal, is just to dump all files into one big directory (4,567 and counting) with elaborate names, then symlink to multiple hierarchies.

It's some work maintain, but so long as music players don't insist on automatically moving files everything functions. Everything is easily findable by simple file name search in the big directory, or by drilling down the hierarchies.

It would be nice if it were easier, and if more meta data were supported than I can cram in to the file name, but nothing else I've tried approaches this for usability. I wish I had better news.

Plasma Active Three released

Posted Oct 17, 2012 14:36 UTC (Wed) by mathstuf (subscriber, #69389) [Link] (4 responses)

Media players shouldn't be mucking with tags. I would drop such media players in a heartbeat if I found them. I typically use MPlayer locally with either MPD streaming or local files. Even if MPD gets the "feature" of retagging, all the music is on a read-only nullfs mount in its jail.

If you have "halfway" tags, I'd recommend using picard to tag your music. Most of the work is in "yeah, that looks good" before applying tags fetched from musicbrainz. If there aren't enough tags filled in, picard can fail to auto-match with the database, at which point it's a fairly manual job[1].

[1]Add a "tport=8000" (at least, port 8000 is the default) query parameter to your musicbrainz URL when picard is running. It will then add links to insert release data straight into picard.

Plasma Active Three released

Posted Oct 17, 2012 15:32 UTC (Wed) by sorpigal (guest, #36106) [Link] (3 responses)

> Media players shouldn't be mucking with tags. I would drop such media players in a heartbeat if I found them
You'd be surprised what "popular" players will do, but note that I include in this players with built in tag editors of highly dubious quality.

> If you have "halfway" tags, I'd recommend using picard to tag your music.
I'll give it another look (it's been a while since I tried one of these), but...

tl;dr auto tagging is broken, manual tagging is necessary for correctness

I find that tag-guessing and batch tagging are dangerous. Such services will happily write incorrect tags, possibly overwriting my manually-chosen correct tags.

Consider this scenario: I have a track ripped from a Gilbert & Sullivan "highlights" disc, but I have manually tagged it to indicate which operetta it was from and in what year it was originally produced. An auto tagger may:

(1) Detect that it is from the album I ripped it from and re-tag it with the name of who compiled it, giving it a title based on how the album named it (e.g. "Pirates of Penzance Major General's Song" instead of the correct "I am the very model of a modern major general"), and setting the year to when the CD was released, or setting the genre to "Comedy" instead of "Light Opera"

(2) Identify the album from which the track *originally* came, copying the metadata from that album instead.

(3) Detect that it is a performance of the song, but incorrectly identify the source album as another performance by other persons, copying the incorrect metadata from that album instead.

The result is that I can't sort G&S songs by operetta, year, etc., and in all probability the artist is now listed as whoever produced the recording only, or a haphazard concatenation of original artists with performing artists, possibly without distinction. In my experience the metadata available from common databases tends to be either inaccurate, excessively verbose or organized according to an unpalatable scheme.

This is when it finds my music at all, which in my experience is not common! In the end it becomes a manual process as I feel obliged to manually verify each choice made by the tagger.

Plasma Active Three released

Posted Oct 17, 2012 15:52 UTC (Wed) by mathstuf (subscriber, #69389) [Link] (1 responses)

Ah. I don't have much classical[1] music, so I tend to not have such problems with my tags.
> In the end it becomes a manual process as I feel obliged to manually verify each choice made by the tagger.

Though picard isn't an "auto tagger" (anymore at least) since it won't write tags without an explicit Ctrl-S to save the (selected) files. Newer picard versions show a diff of the tags in the bottom where you can manually fix things (it's not a full tag editor, so adding new tags isn't possible with it AFAICT), so detecting "picard did it wrong" should be fairly easy with it.

[1]I would expect that simiar issues arise with things such as opera (as you point out), theater, broadway, national anthems, and so on. I guess "performance" pieces is a better name for this type of thing?

Plasma Active Three released

Posted Oct 17, 2012 17:32 UTC (Wed) by sorpigal (guest, #36106) [Link]

> I guess "performance" pieces is a better name for this type of thing?
id3v2 does specify fields for composer, lyricist, original release year, etc.., but many of these are ignored or silently discarded (!) by taggers/players. Ogg comments are just unsuitable for this level of detail, because there's little agreement on what fields can be expected to exist beyond a very basic, limited set.

Plasma Active Three released

Posted Oct 17, 2012 16:52 UTC (Wed) by intgr (subscriber, #39733) [Link]

> tl;dr auto tagging is broken, manual tagging is necessary for correctness

Well, ideally, you would enter all relevant information into a global database and simply configure your tagger to store the tags based on your preferences; that way you don't have to do redundant manual work and benefit from data that others have entered.

> I have manually tagged it to indicate which operetta it was from and in what year it was originally produced
> giving it a title based on how the album named it (e.g. "Pirates of Penzance Major General's Song" instead of the correct "I am the very model of a modern major general"), and setting the year to when the CD was released

Of course people have a different understanding of "correctness". The MusicBrainz schema is fairly flexible, there are three levels of detail:
1. Track as labeled on an album track list
2. Recording, which links all different releases/tracks of the same recording
3. Optionally, there's "work", which collects together recordings of a single work, but can also relate to other works such as the opera.

Sounds like what you want is to actually store the attributes (title and date) of the work (#3) in your tags, not the track list (#1). Now, Picard itself doesn't support this, but my point is that MusicBrainz isn't like the CDDB databases that only knew about track lists. MusicBrainz a first-class music database and the community actually cares about correctness. With some additional code it would be possible to fetch all the relevant details you care about.

Not all editors enter information at this detail -- and earlier versions didn't have all these capabilities. But that's a small matter of going in and entering/fixing the data, which you already have to do when building your directory structure anyway.

> or setting the genre to "Comedy" instead of "Light Opera"

MusicBrainz doesn't support genres as such because (as you point out) people often have wildly different understanding of genres. But it does provide a "folksonomy tags" feature. Picard can be configured to only accept genres that you personally have entered, or set a threshold on how popular the tag has to be to be applied.