#musicbrainz

<03nik> Anyone had success of making a postgres docker image with a snapshot of the musicbrainz dataset loaded onto it?

<04elomatreb> @nik: you can use https://github.com/metabrainz/musicbrainz-docker to get a container like that

what should i do with this work? https://musicbrainz.org/work/fc5dc54a-f795-35ab... - it's only referenced by one recording, and i'm replacing the 'recording of' relationships on that recording with the individual tunes... unless i should do something else? i'm not sure there's much benefit to having a work for a specific set that was only recorded once

(because that would end up creating many thousands of single-use works that don't provide any information beyond what the medley-of works already have)

lw (IRC): remove the relationships, and it will eventually be autoremoved

thanks

although now i'm wondering if it would be worth creating works for these just in case someone else recorded the same set. at least i'd consider that for some more common sets, not sure about these ones...

If it's very commonly recorded as the same medley, sure, but I'd expect that'd be rare that a specific medley is that common?

I guess if an artist makes a medley, it becomes really popular and then other people explicitly play their medley?

well, for example some of Michael Coleman's sets have become fairly common just because he was an influential early recording artist in that genre

That seems like a case where they might be worthy works of their own then

i'm pretty sure that's not the case for this example though :-)

As a person who knows the genre well, you're probably more qualified to decide on where that could make sense :)

Yeah, tune sets should mostly just go away as works. Coleman’s and others may be common, but even they thought of them (I am pretty sure) as assemblages of distinct works.

I can see an assemblage becoming a standard worthy of a work, even if it wasn't originally seen as that

But yeah, sounds like it should be rare

right, i doubt Coleman intended for people 80 years later to be posting on thesession to tell people they can only play a tune in the same set Coleman played it in (a comment i've actually seen)

huh, i was about to merge two recordings but it turns out they actually put a shorter version of the song on the CD than on the vinyl. that's weird, it's usually the other way around

i really hate this trend of using gatefolds for CDs, it's impossible to get the damn thing out of the packaging without scratching it

shouldn't "banjo" (https://musicbrainz.org/instrument/6bf88fc7-a23...) be a family, not an instrument? "banjo" by itself is not a specific instrument

or else "banjo" should be the 5-string and tenor banjo should be a derived instrument

picard show multiple value for ISRC (TSRC for id3v2.4), but why `ffmpeg -i filename` just show 1 value for that tag?

i know absolutely nothing about US geography, if an album is listed as being recorded at a studio in Nashville, TN, is it possible it was actually recorded at a studio in Franklin, TN?

feye: ffmpeg doesn't properly support tags with multiple values

kepstin Ah, I see

Anonther question, what id3 version you use? And why v2.3 can't use utf-8? And why default v2.4 use utf-8?

ID3v2.3 can't use UTF-8 because the spec says it uses UTF-16. they changed it for v2.4 because UTF-16 is basically a dead encoding (except for legacy use on Windows / Java)

most of my files aren't mp3s, but I use id3v2.4 for the mp3s that i do have (all my players support it fine)

Isn't utf-16 more char than utf-8?

it depends on what script you're encoding, but for Latin scripts yes, UTF-16 takes up a lot more space

for text using latin alphabet, utf-16 usually takes up more space than utf-8, yeah.

(which is probably partly why v2.4 switched to UTF-8)

for chinese/japanese/korean text, utf-16 is usually smaller

then v2.4 is good, why v2.3 still popular?

mostly that there's a bunch of old tools and players around that people still use which never got support for 2.4

because v2.4 is newer, so a lot of existing MP3s were tagged with v2.3 when they were ripped, some software still doesn't support v2.4 at all

most (or at least many) tagging apps can replace v2.3 with v2.4 when you re-tag a file, although i don't know if Picard does that off hand

in practice there's no real benefit to using v2.4 unless you need the new tag types introduced in that version, v2.3 is more compatible and the difference in space usage is irrelevant when an MP3 is several megabytes

picard will convert id3 to the version selected in the options when saving a file if changes were made. not sure if it changes that if you re-save without any changes.

i mean, i'd use v2.4 for new tags but i wouldn't put any effort into converting old tags unless i was editing them anyway for some reason

(i think TSST (set subtitle) is new in v2.4? i remember that being an issue when tagging MP3s)

in general, i'd say use v2.4 unless you run into an issue with a specific player not supporting them - in which case, try 2.3 instead.

v2.4 published in 2000 based on wikipedia and still some player not support, that's suprising

Oh wait ffmpeg can't even read multi value tag from id3v2.4

98% of people who listen to MP3s (not in this channel, obviously) don't care about tags except for artist, title and album. so there's little motivation for software to add support, outside of software intended for serious music collectors which mostly does support v2.4 now

i mean, iTunes still doesn't support FLAC, which should tell you something about the state of mainstream music playing software :-)

Because they have ALAC I think

yeah, but what i mean is, FLAC is by far the most popular format for lossless music collections, and yet one of the most popular music players doesn't support it... because those apps are not aimed at people like us

(iTunes and Windows 10 both apparently support ID3v2.4 though, so that's something)

That's great, but I use linux :p

i would be surprised if many (or any) open source players didn't support 2.4 nowadays

unless they're like, unmaintained for the last 20 years :-)

As always

lw: Franklin is a suburb of Nashville, a ½-hour drive. Definitely possible that a “Nashville studio” is actually in Franklin, but also possible that one studio business has two locations, or that the studio business moved locations, or that there are two completely separate businesses using similar names…

yeah, hm

it's for this release: https://musicbrainz.org/release/3d118cae-dede-4... - the liner notes say "Recorded at The Playground in Nashville, TN", while MB has "The Playground" in Franklin and "Playground Recording" in Nashville. so i put it as the Franklin studio, but i'm wonder if that's wrong... or at least not certain enough to add to the database

annoying that none of these big names ever have an email address on their website, or even a contact form

always write-only, "Join our fan club / mailing list"...

lw: Discogs, FWIW, says that album was recorded at https://www.discogs.com/label/268044-Playground...

“Sometimes credited as being in Franklin, TN.”

Not very helpful…

huh

maybe two studios run by the same company? i'm not sure if they went out of business, but the website doesn't work anymore :(

i feel like based on that i should update this album to the Nashville entity though, before i added these relationships the Franklin one only had a single recording

then again the Nashville one only has a single recording too

does no one record anything there? that might explain why they went out of business :-)

Discogs lists a bunch of stuff recorded there.

https://soundville.us/ is now at the Nashville address but seems unrelated

Mettea recorded a few albums with Playground, but it does seem like they’re out of business; only old articles can be found now.