MeB summit 2020: MB breakout session about classical music?

Tags: #<Tag:0x00007fe30caf1468> #<Tag:0x00007fe30caf0fb8>

Hi everyone,

I was wondering if we should have a session about classical music in the current MeB summit

(link for the people unaware: https://wiki.musicbrainz.org/MusicBrainz_Summit/20)

I see as possible relevant topics:

  • future community cleanups: most composers-as-recording-artists are now in a good shape (meaning at most 2000-3000 recordings instead of 10,000+) but still need some effort. Or do we have cleanup topics? (minor composers, performers, instruments, labels…)
  • do people use events for classical? What is missing to make it more useful?
  • opera (the-topic-that-must-not-be-named): are we ready to edit operas efficiently?

If there’s interest then we should plan for a session tomorrow IMO

One that needs a cleanup is Haydn, I wouldn’t count him as a minor composer. I don’t use events, but that’s just because that’s not the type of data I focus on putting on MB (namely, performers, works, composers and performance dates to spot dupes easier). There’s definitely room for improvement for the UI of MB (I dream of an option to restrict work search to the track artist).
I was unaware of the summit, if there’s a section I’ll tune in, even if just out of curiosity.
Dorian

1 Like

I don’t really know how events are used normally, I only look at classical music, but for me events make sense only in very few cases: for live recordings at some specific occasion (for example if someone recorded Rostropovich at the Berlin wall) and maybe for premieres. That covers only a very tiny fraction of classical recordings. To make an event is to link recordings to something bigger, that only makes sense if something bigger is actually happening.

Just wondering why opera would be “a topic-that-must-not-be-named”. I’ve been rather busy on opera for the last few month, supported (informally) in this by several other editors, without major problems. Verdi is now down to zero composers-as-recording-artists, so are (more or less) Donizetti, Offenbach, Humperdinck. Now working on Bellini and Rossini before going for the Frenchies (Bizet, Massenet, Gounod, Grétry …) Of course having some help (and some more tools) would be nice, but corona gives me a lot of time to work on these :wink:

As for Events I’m with @Dosoe : Not really of primary importance for classical

In the artists I follow, Tchaikowsky, Schumann, Mozart and Haydn still have > 3000 recordings and would enjoy a community cleanup. Vivaldi and Haendel I don’t follow and are probably above.
Still that’s tremendously more manageable than what we had a few years ago with > 10,000 recordings for most of those composers (Mozart was at > 50k recordings in 2011).

You will have to ask more veteran editors than me, but I think that’s a guideline topic that came recurrently with no progress (how to split acts and so on).

That’s huge news! I don’t follow Italian composers so I didn’t know. :muscle: :pray:

I was mentioning operas because that’s IMHO the most difficult case to enter correctly (finding the right work, adding the right singers and their role on each track, etc.) even if userscripts help. I’m pretty sure we have ways to improve the UI there, and that’s where a summit session could help

One topic I have recurrent trouble with is works that have no definitive ordering or that exist in different versions, one example that springs to my mind is Rameau and his Indes Galantes, they exist as suites, as an opera, it feels like they are never split two times the same way and people love to record chunks of it, works are hard to identify and I have a hard time editing recordings of it. Also, the water music by Händel has its own thread in this forum because of that.

One more ambitious thing would be the following: As acoustids basically use the notes that are played (in the way they handle the spectrograms), could there be a way to somehow detect the works using acoustids? Or at least make propositions? I know that’s much more complicated to implement, but it applies more to classical music where the same works are recorded again and again. It would require the complete acoustid (i. e. on the whole recording, not just the first 2 minutes that were used initially if I remember correctly) but we have quite a lot of those.

Interesting question, I guess acoustID does not have enough data for this but AcousticBrainz might have? Pinging @alastairp

To detect the work automatically is a tedious task and still an open subject in the scientific community. You can see a recent effort done here: https://arxiv.org/abs/2009.10128

1 Like

Sorry, AcousticBrainz doesn’t have enough detail to determine if different recordings are of the same work.

I think events are really interesting because in many cases works that have never been recorded have been performed at events (premiered, at the very least). And well, I see no reason not to track what’s been played at the Proms or the local symphony! But it’s certainly not a big priority except if we want to try and compete with Bachtrack.

That was me :smiley: The old opera guidelines were super strict, and the current ones are sometimes confusing. I actually feel they work reasonably fine, but then, I don’t fix Wagner. Operas do require a lot more effort than other works to get right (as in, you need to figure out the numbers for each act, then figure out how they match the sometimes entirely random-seeming track breaks), but it’s probably not as bad as that made it sound :slight_smile:

I’ve been fixing up some Schumann recently - a lot of that seems to be a couple huge boxsets, so if we found the metadata for that and the time to fix it we could make big progress there :slight_smile: