#musicbrainz

:)

CatQuest, i actually do happen to agree with the style guide in this case

I on the other side was a bit unaware of that guideline change, if the underlying tracks where infct real mashups and if this was a ganhe to the tracklsit or what

like i said, i have disagreed with it in the past

I make change following sryleguidelines if I don't agree with them, if they make "sense" still

even if i don't agree*

i need ot eat and its late

hmn

i mean I remember styleguide disaster no 5

i hated that

but alas I had to follow it

which was that?

oh that

uuuhh

pre-ngs an pre advanced relationships

it was in 2006 or something

awesome

oh it should have been Max_Gentleman

what was the guideline?

okay, go have your... dinner? good chat, CatQuest

CallerNo6: I assume the principle is that the release in MB should reflect the release in real life…certain formatting and correction aside.

kepstin-laptop, do you use opus on mobile?

I use opus on my laptop and on my ipod, which runs the rockbox open-source firmware

heh, havent used rockbox in ages

they added 2048 as a game recently.

the "iPod Classic" that I have isn't one of their officially supported platforms yet, but that's only because they don't have an automatic installer yet

and nobody's figured out how to make it dualboot the original firmware either, it's a replacement only thing.

i need a new battery for my ipod classic

it's oold

just using my smartphone for that stuff now

heh. the replacement batteries are pretty readily available online

although I spend a lot of time in wifi range, and could use music streaming then

hmm.... $43 for the drive

some people have actually replaced the drive in these ipod classics with modern msata SSDs

can get up to 512gb

(at some battery life cost)

wow, that's really nice

why opus instead of stuff that gets better battery life?

or does opus get good battery life now

Because Opus.

opus lets me fit more music on the drive, mostly

it's still more cpu hungry than other formats

rockbox folks have been steadily optimizing it; a few mhz here and there

hawke1: information always needs to be parsed, whether we realize we're doing it or not. and when we don't realize we're doing it, we're probably making assumptions.

So...I'm exploring relationships and I cannot find any documentation on the various types, or what 'direction' means, etc. Where is this documentation? :D

I think my ipod can decode most of my music at an (average) 60-70mhz now

with mp3 and vorbis, rockbox runs at the minimum processor speed of 54mhz constantly tho

_Dave_: https://beta.musicbrainz.org/relationships

"beta"? ;)

oh, should be the same, beta or non-beta

is opus better than aac now?

it is...I wonder why google did not find this page?

_Dave_: direction only applies to some relationships. For example, if I'm your sibling, you're my sibling. direction is irrelevant.

but if I'm your parent, you're probably not my parent.

In the second case, direction matters.

yoasif: the open-source opus encoder produces way better quality than any open-source aac encoder. for what that's worth.

I see

kepstin-laptop: not that that says much, since the open-source AAC encoders are kinda shit by my understanding

lol

encoding data like that is bound to be missed by some programmer ;)

yoasif: in some lowish bitrate (64kbit?) comparisons between opus and a good aac encoder (nero?), I think opus came out a tiny bit ahead in most samples

at more useful cd-transparent bitrates, i doubt you could tell them apart.

but the fun thing is that opus can do this with only ~25ms of delay, which is good enough for a phone call. And if you push the bitrate up a bit, that can go below 10ms.

yeah, i don't mind using free (non open source) encoders

opus was originally designed for realtime stuff, the fact that it's also good for general music is just a bonus.

http://listening-tests.hydrogenaud.io/igorc/res... opus beats apple aac in 64kbit

ah, yeah, that's the one I was remembering. so it was apple's encoder, not nero.

_Dave_: Okay, I'll rephrase. Each relationship has a direction, and can be "reversed". So e.g. [recording b]<is a remix of>[recording a] is one direction. The reverse, also true, is [recording a]<has remix>[recording b].

oh, they had both

I guess there's not enough information on the relationship documentation page to enable me to parse this "relations" array from a JSON query ... no doc on 'type-id' for example

for reference, I'm trying to get at the work through the recording

I just realized you are probably handling medleys this way ;)

hmm? that shouldn't be too bad to do

the only issue with the webservice responses is that relationship attributes aren't represented well.

It isn't in the general case. It's the edge cases I'm concerned about (and as most of you are acutely aware, this is where most of the work goes)

yeah, apple has beat nero for years now... was a pain to use it without quicktime installed (on a linux system ;))

_Dave_: medleys are a bit special, since we're currently in the middle of switching from a separate medley relationship to an attribute on the standard recording-work relationship

but for example, i'm parsing http://mbjs.kepstin.ca/?release=bef90838-a09d-4... out of the json webservice

(that website is a json webservice frontend, but rather incomplete)

and it didn't give me json

or anything really

am I supposed to type stuff in the search box? :D

hmm? if you have javascript enabled in a reasonably modern browser, that should show a web page for a japanese album

it doesnt, but I'm not allowing twitter code to run in my browser either

bah, I might have a bug that makes it fail if the twitter doesn't load

I love noscript :)

which would be strange, since I'm not even using twitter on that page

-why- do we need twitter code?

I dont know

kepstin-laptop, that is super fast

the type-id matches the id in the work array

hmm? this is my personal page, nothing to do with musicbrainz. But I use twitter code to display twitter links on artists with such.

so there's a relations array, and I'm assuming that could have many hashes inside it

yoasif: I'm running a clever caching proxy in front of the musicbrainz webservice to speed it up. But results can be up to 24h old.

each of these hashes has a 'type-id' (no clue what that means really)

and then there's what I want, a work key in the hash

_Dave_: I mean for parsing the json (or xml), the type-id corresponds to the type, which is also given.

the type ID corresponds to the UUID of the relationship type, which you can get from the relationships list page.

you go through the relationships array looking for the one with the uuid of the rel type you're trying to find

so if I did a ?inc=work-rel I should only expect to see works in the relationship array?

yep.

ok so...what if there is more than one?

then the recording might be linked to more than one work

for example, if it's a medley

wat

_Dave_: http://musicbrainz.org/ws/2/recording/b3d51fe2-... is an example of a medley, all of the link works are songs included in the medley

is it lame because there is a depedency? :D

I can't read japanese LOL

but I see what you are saying

notice the UUID in the type-id field, that corresponds to https://musicbrainz.org/relationship/a3005666-a...

which tells you what the relationship means

but..it's the same as the work id in my case

does this mean "it's a related work"?

that shouldn't happen - you have an example query?

the "type-id" is the UUID of the relationship type, it has nothing to do with any entities in musicbrainz

the entity id will always be in e.g. relationship['work']['id']

for the related work

ok give me a moment to spit this out in my code

so "relations" is an array, each array entry is a "relation". A relation has a "type-id", which tells you how to interpret the other contents.

in this case, the type id "a3005666-a872-32c3-ad06-98af558e99b0" means "this is a link between a recording and a work, so there's a work entry here"

then you look in the work entry, and get the work id

which takes you to http://musicbrainz.org/work/aef39ce6-69b9-4184-...

?

(and also has some stuff like the title, language, type, etc)

so that id is not a work id?

arh

the type-id is a type id

the id in the work is the work id

the work id here is "aef39ce6-69b9-4184-bdf8-a92d7d8eec3b", because that's the "id" field in the "work"

ok for some reason those IDs weren't different for the query I was looking at

but when I access now it works

you might see some recording-work relations with type-id="12ac9db0-ec26-3567-be3a-2e662e617803", these are old medley ones.

I'll put some code in to detect this condition at any rate

but most should be "a3005666-a872-32c3-ad06-98af558e99b0"

all this to get the song title

what are you starting with? a release? a search query?

Thanks, btw, for your help :D

you can do a query to get works directly from the release query, if that's what you have. Saves a bunch of steps.

an arbitrarily encoded music file

through acoustid

gives me a recording