#musicbrainz

Soul Food

(full stop)

I'd use either 'Soul Food' or 'Soul Food: Music from the "Soul Food" Motion Picture'

you guys illustrate my hesitation :)

hehe

I'll just leave it as it is I guess

I agree with navap.

Lotheric: I am disappointed that that's not a link to a CAA image though. ;-)

it is now, I uploaded it after ;)

I figured it would be soon. :-D

If it helps, I prefer just "Soul Food" because I find the rest sort of "meh"...

I'm not a fan of using "Original Soundtrack" subtitles on other soundtrack releases

me too

I'd use the rest (or disambiguation) if there was also a score

but I took a look at the soundtrack guideline and I was confused by the cover

what is extra information and not part of the title ?

It's a terribly inconsistent cover. I just saw that they further confuse things by stickig "Soudtrack" as a subtitle on the front cover and disc

JonnyJD (and jesus2099): I usually do lxterm -> tmux -> ssh/mosh :p

Freso: Jam Master Jay “ended” when person performing as Jam Master Jay died. I don’t know that peformance name, but I doubt it started at the same time he died…

JonnyJD luks: Wouldn't it be possible to read the audio straight from the CD without ripping it per se? E.g., like CD audio players do?

theoretically there could be one performance name, used by different people over the years… which would make there only one begin date, different perfoms as date, and no birth=start date would make sense there

Freso: if acoustid uses only streaming algorithms, yes

SultS: But Jam Master Jay is still an artist:Person, for which start and end date means "born" and "died".

SultS: If you want to introduce another artist: type with other semantics for start and end (e.g., like group and character), then go ahead.

but it won't actually be faster than ripping

JonnyJD: Alright.

Freso: but it conflicts with *begin* and *end* area, which are used together with birth and death date

SultS: Sure. So think up something clever and submit a -style RFC for it.

The current guideline in that respect is from before we had Areas.

yeah

I’m not certain we need new artist type though

oh, well, I’ll think about it some more tomorrow… bedtime for now

I'm tempted to add this to the database: https://www.youtube.com/watch?v=NadyUs_DWK8

should I?

Well I can't see why not I guess

JonnyJD: the audio processing in chromaprint is just streaming

there is a bit of post-processing that is done on the whole thing, but you don't need the complete audio at once

in fact I often do silly things like `curl http://xxx | fpcalc -` :)

drsaunde: https://www.youtube.com/watch?v=zv9A9K4YcHI

luks: do you have to give the file format somewhere? I can't get it to work with cdparanoia: http://pastebin.com/JEJz43Jq

if I read http://musicbrainz.org/doc/Style/Recording#Live... correctly, any "(live)" should be removed from the title, and disambiguation should be used instead?

does it matter whether or not the release include "(live)" in the track listing, or does that only affect the track entry itself?

jonny, based on the error mesage, fpcalc expects to be able to find the track length at the start; so you can't use it as a filter it seems

cdparanoia 1 - > tmp.wav; fpcalc tmp.wav; rm tmp.wav would work, but requires disk space

well, luks said he does `curl ... | fpcalc -` and it works

hmm

may then depend on how curl and cdparanoia write to the output

`cat cdda.wav | fpcalc -` also doesn't work.

that's very odd then

and "file cdda.wav" reports it as valid WAV data?

yes, I can also listen to it fine

(with sox/play)

and it's regular 16-bit 44.1kHz stereo? sounds like a bug in fpcalc then

cdda.wav: RIFF (little-endian) data, WAVE audio, Microsoft PCM, 16 bit, stereo 44100 Hz

not sure if it's a bug. Some minutes ago I didn't even know fingerprinting from a pipe is supposed to work

yes, but if it also doesn't work on an oherwise valid WAV file...

maybe it nearly works, or some format need a little more input. possibly I can prepend the length somehow

no, it works like `fpcalc cdda.wav`

just not over a pipe

ah ok

well, curl could be waiting until the file is transferred (to memory or a temp file) before sending it to its output; in that case, fpcalc's input processing may work "by accident"

well, at least not in my case `curl http://kraehen.org/tmp/cdda.wav | fpcalc -` also fails both (fpcalc with the same error and curl with some mismatch Oo)

http://pastebin.com/ct6gTzKS curl: (23) Failed writing body (1157 != 1440) possibly my fault, because I didn't make sure the file type is given correctly

I'm having some really frustrating results attempting to build the musicbrainz server inside Docker

The command that installs the perl deps fills up / with all kinds of garbage, and subsequent commands don't actually work.

ah okay, the curl/wget error is probably just because fpcalc closes before curl/wget can push the whole file to the pipe

Is this behaviour something someone else has wrestled with, or should I ask somewhere else?

I've already asked #perl, they couldn't figure it out.

blast_hardcheese: #musicbrainz-devel is generally more suited for these questions.

Freso: Ok, thanks

However, it's also the weekend (and in the middle of the night for the main developers), so answers may still be slow.

Well, it is midnight for me as well, I should probably sleep ;)

OTOH, it's the middle of the night for the main devs, so they may very well be around. :p

I was really hoping to get this built today, this is day three of debugging.

when I encode a flac from cdda.wav `cat cdda.flac | fpcalc -` works. So it just doesn't work with wav and needs to know the length/duration of the wav somehow

for anybody still interested: `cdparanoia 1 - | flac - | fpcalc -` works, too

JonnyJD: Working on the CD AcoustID submitter? :)

wouldn't say "working" more like playing with the basica around it. I was quite interested when luks mentioned calculating from pipe already works ;-)

I do have another algorithm to implement this weekend..

:)

Freso: an idea where I should put this informatin about piping from CD?

Uh. The wiki? :/

it really sounds like "acoustidsubmit" can work, but I won't work at it that soon

Submit a patch for fpcalc's README?

hm, not sure if its "good enough" to commit already

and wiki: doc/AcoustID?

I've no idea. :)

if somebody feels like blogging this, that would also be a start (and make somebody implement something) I can't prepare a full blog post right now

maybe I do it exactly the same I did with isrcsubmit: start with posting command lines in the forum

ah and Lotheric: ^ (getting fingerprints directly from CD is possible)

`cdparanoia 1 - | flac - | fpcalc -`, probably needs more tweaking, but works (track 1 that is)

first thing to change would probably lowering the "paranoia level" to make reading faster. We don't need to make sure every bit is as on disc

I can't even find where pyacoustid is linked (python web service wrapper)

sampsyo: ^ (maybe you are interested in this, since a acoustdicsubmit (from CD) would probably use your library) Or maybe your tool already can do that and I just didn't know

you need to make further edits

Ok, cheers.

Freso, luks, Lotheric, sampsyo: I created a forum post. Please add to that if you have additional information (or see additional problems): http://forums.musicbrainz.org/viewtopic.php?pid...

Maybe some student is bored in between semesters and makes something like it

i'm going to have a quick nose around in the fpcalc sources. I think it should be possible to adjust it to handle being run as a filter

incidentally jonny, cdparanoia also supports ripping to aiff/aifc instead of wav - does that help fpcalc?

Zastai: cdparanoia -f and -a work, thanks for the tip