#musicbrainz

computerfan: It will only be able to tell if the two sources contain audio tracks that are different enough for the fingerprinter to give them different fingerprints.

Ah, Picard would probably be using Acoustid to scan the files, so it'd be quite unlikely it could distinguish a difference.

The cover art is manually linked to a release, which in turn is linked to recordings which can have Acoustids attached.

other than videos often have more silence or near-silence before the song starts

which could result in a different acoustid IIRC

however mostly it depends on the linking acoustid <-> recording, so :)

ianmcorvidae: That also depends, I think Acoustid doesn't pay too much attention to leading/trailing silence.

I've seen problems reported before with it dealing poorly with leading silence

trailing won't matter unless the song is very short, since acoustid uses only the first two minutes at present

CatCat: I don't think I can put up the dough to travel, even if it would be nice. :/

Ah, good to know. Funny, though (sort've related)...I've scanned two files with different fade-in audio and they generated the same ID.

Freso: you're in Europe though, yes? There will be accommodations provided, and in the past there have sometimes been partial sponsorships for travel costs

yeah, it's all thresholds; hard to say exactly what will and won't generate new acoustids without, well, trying it :)

considering as I get my music from multiple sources I don't want take a chance of data being uploadedthat could define whether or not the music was obtained through youtube or through Amazon or iTunes

I'm kind of security paranoid

ianmcorvidae: Hohum.

computerfan: nothing will get uploaded with your name attached to it unless you hit the 'submit' button, in any case

ianmcorvidae: I'll think about it.

ianmcorvidae: And yes, I'm in Europe.

computerfan: the scan button is just plain lookups, nothing identifying attached except perhaps an IP, and that's only going to acoustid itself

thank you for all your help and by the way good luck on your trip

and it's just sending a fingerprint anyway :)

thanks

Freso: yeah, so you'll have a comparatively cheap travel budget, anyway :)

ianmcorvidae: I added me to the list for now. We'll see what happens. Feel free to poke me with updates as I suspect I'm going to be hellishly busy from next Monday and until year's end, and won't be able to keep much up with anything. :)

(I mean, that /is/ the name he performs under)

CallerNo6: They're not mutually exclusive.

maybe I should have asked on [mb-pedantry]

which one? :) (I'm guessing o'hare)

hi ian

are you one ofd the developers

He is but he's not here just now

(will be back in a few hours I guess)

are you

Nope

im looking for an example of a audio fingerprint and the data that they contain

fingerprints are just unique id's, kind of lime MBIDs

s/lime/like/

unique to a particular song. so if you you buy an mp3 on amazon, and I buy the same song as an .m4a on itunes. they should have the same fingerprint.

not necessarily

but usually

_should_ :)

reosarevok: FTL is fantastic, by the way.

:)

Good!

demosdemon: Fingerprints are not unique IDs; AcoustIDs are.

computerfan: You can find a visualization of the fingerprint by going to any acoustID page and clicking on one of the numbered fingerprints

Also, the output of 'fpcalc' is the fingerprint data, which is then looked up to find an AcoustID.

yeah that, that's what I meant <.< >.>

reosarevok: Also, cross-platform. Yay!

Yeah

I am tempted to buy it

But I have a huge backlog

So I think I'll buy it for myself for Christmas

hehe

so no metadata is uploaded when i hit scan?

Not when you hit scan, no.

It basically takes the output of fpcalc and puts it into http://api.acoustid.org/v2/lookup?client=8XaBEL...

With X of course being the duration and Y being the fingerprint itself

this technology is amazing to me

FTL is awesome, I backed it on kickstarter

New post: blog: Possible date change for our summit: 9-11 November 2012 <http://blog.musicbrainz.org/?p=1589>;

smgoller: It needs a few things, but I expect they’ll be coming and/or in a sequel/expansion

(mainly a sandbox mode)

you mean without the rebel pressure? :)

Yes

And the limited sector count

yeah, make it an unlockable

I’d rather have it as an alternate mode, but unlockable would be OK with the right unlock conditions.

Apparently you can easily mod it to disable the rebels though, so at least there’s that

Relevant to discussion of performance names / artists with multiple names? http://tvtropes.org/pmwiki/pmwiki.php/Main/Alte...

andrew dice clay comes to mind

And mentioned on that page.

oh. ha. hadn't read that far.

I meant mostly the intro there — is there some point between type 1/2/3 there that we would want to have separate pages for MB artists that follow those descriptions.

i.e. are those 'type' guidelines relevant to our case at all

I /had/ wondered about eminem/slim_shady. It's a different "persona" but eminem is still credited (maybe just for marketing purposes, I dunno).

Hey. That wiki page links to itself.

Hmm I've wondered about Ziggy Stardust and The Thin White Duke too. I mean, it's interesting but does it fit into musicbrainz?

Dunno.

It seems a bit odd that we are so inconsistent in that regard — sometimes artists are on a single page (with a nod and a wink given to the guidelines), and sometimes they’re completely partitioned.

agreed. I think we're stuck between "an [artist] is an artist" (meatbag) and "an [artist] is the name that appears in a credit".

ACs and aliases brought us closer to the former, but not all the way.

So what’s missing? Is it guidelines about when to use one or the other? Or is there some feature that would make it work in the data too?

hmm. I think we don't really have a "real person" entity distinct from a "persona" entity.

Sure we do, we just call them both 'artists' and relate them with 'performs as' ARs. I think that’s pretty much all it would be anyway

But not always. "Should I go ahead and create another entry for the LegalName as well? -- Usually, no"

(from http://wiki.musicbrainz.org/Style/Artist/With_m... of course)

That only applies when there is only one persona though

Who can say when it's a different persona?

The artist, presumably. :-)

Also, all the ways in which “personas” are relevant it seems to me are taken up by Artist Credits.

(some others disagree, of course)

I'm finding some editors want to limit the use of ACs.

... because music players can't make use of them, and treat it as a different artist.

Seems like a problem for the tagger, though.

that is, because it messes up their preferred tagging system

Not for the underlying data.

Yeah, totally.

But how far can we push the use of ACs when the tagging side of it hasn't caught up?

I would say push it as far as we think is correct (that is, get the data as correct as possible; not use ACs as much as possible).

Until there is data to use, the tagger apps have no incentive to catch up

I mean, how can you even test if tagging will turn out like you want, without knowing what the source data will look like, and without a good sense of what you want it to look like in the first place.

hawke_1 says, " all the ways in which �personas� are relevant it seems to me are taken up by Artist Credits." Is that a departure from rfc 149, or am I reading it wrong?

It is, because RFC-149 is intended to be more of a compromise that what I would personally prefer

Okay, right, which is partly what I mean by "stuck in the middle".

and probably what you mean too by "inconsistent"

Yep.

Actually by inconsistent I meant more that the guidelines aren’t always followed and seem to be based on twisting to fit one person’t perspective.

(The “but what if” questions in the Q/A format)

So. We'd have to be able to make a case for "performance ACs are factual, not 'as credited' "

Yep.

That's what I get for bringing up DJ&TLT