#musicbrainz

http://musicbrainz.org/showmod.html?modid=4735220 yes votes/approval please?

is there no accoustic fingerprinting/analysis process that is fully open sourced? From reading the wiki I have gathered that generating PUIDs is completely closed source process?

MusicDNS do all the PUID generating. thats closed source

mudcrow, There is no open source alternative for such a need?

i'm confused

musicdns.org talks about open source

and musicbrainz

the analysing part is closed source.

I dont know of any other that is open source, ruaok is the one to ask about that

odd

so there's a client library

and it talks to a closed server?

Apparently TRM was a closed source processa lso.

*also

This FDMF is apparently open source.

malverian: i did some tests with FDMF myself

the false positive rate was ridiculous

Hmm...

The fingerprinting idea is such a good one..

I just hate to think it's bound to some closed source server staying "free"

thanks mud

no prob. gurt

there's a whole bunch of gpl source in this musicdns lib

i'm trying to figure out where it talks to something proprietary

flamingcow, Well the closed source portion is apparently the part where it talks to the server and says "generate a PUID from this music signature"

ah, ok

same degree of lame

Yeah.

It seems like an unacceptable sacrifice..

so write something better

When FDMF could be improved instead.

flamingcow, Did you mess around at all with the fuzziness options in FDMF?

i honestly don't remember

i really should've written it up

if you'd like to prep a test plan

i have an enormous music corpus to run it over

i don't understand the willingness to use inferior software simply because it's free

because you can improve it

what's the problem with something that's closed source, or partly closed source, if it's the best thing going

ojnkpjg, Because at any moment musicdns could close their servers to us and we would be left with nothing.

instead of getting stuck with trm

no, they couldn't

you didn't read the agreement

and having to throw out everything and having to start over

ojnkpjg, There are many reasons. If you really want to know what they are, read up on the many pages that explain the benefits of open source.

i understand open source

i work for red hat

so, you don't understand open source

you work for redhat :-P

i'm not an ideological nut

ojnkpjg, So what stops MusicDNS from going out of business and their servers going offline?

if the best tool for the job isn't open source, either write something bettter, or quit whining

ojnkpjg, I think you're not being considerate of the potential for disaster in putting all your eggs in one basket.

ojnkpjg, I think we're being perfectly constructive by discussing other alternatives.

the general sentiment that everything that's not open source is evil is idiotic

ojnkpjg, I didn't say it was evil. I said it is an unacceptable risk.

did you even read the agreement mb and musicip have?

ojnkpjg, Do you understand that MusicIP could go out of business?

ojnkpjg, You have refused to acknowledge that concern.

Little paper agreements do not magically save a company from bankrupcy.

If the server costs become too much, etc.

flamingcow, Anyway. There is a good wiki article that I linked earlier where the author seemed to have good results with FDMF

right

flamingcow, I'm considering emailing the author of the program and asking him what he thinks of his algorythm working with MB.

and i'm trying to figure out why mine so blatantly disagree

i was working with 20k lossless files

and had thousands of false "matches"

flamingcow, Hmm.. Perhaps something to do with the fuzziness that is used by default. Were you using an older version perhaps?

this was more than 6 months ago, so almost certainly :)

flamingcow, I will try some of my own tests. I have about 12,000 lossy files that I could test on..

Probably 200-300 duplicates.

i'll rerun with the current version

* Other projects that wish to integrate fingerprinting services into their applications will need to sign up with the MusicDNS service. This service is free for non-profit projects (musicdns.org), and price-tiered for commercial projects such that even small startups have access. For more details, please visit MusicDNS.org.

* $10,000 held in escrow for MusicBrainz, plus contractual commitment to supply hardware resources should MusicIP exit the fingerprinting business. This is designed to allow us to continue the service should they decide to stop providing the service.

* a 10% cut on all income earned from fingerprinting queries where MusicBrainz metadata is provided via the MusicDNS service.

* 800K acoustic fingerprint ids (PUIDs) that which are already loaded into our DB.

hmm.. i click on edit all for an album, paste a list of track times into the parser, parse the times, click continue, and then it say there were no changes made

check the times are in the correct format 1:30 and not 1.30 or 1,30

ojnkpjg, Interesting.

yes they are mudcrow

is it because the album has a discID?

ah yeas, you cant enter times if it has a DiscID

is there a contractual guarantee that the access for nonprofits will always be free?

ah.. the warning/error message dosen't say that

"Adding, removing and renumbering of tracks disabled because of valid Disc ID(s)"

(and track times)

ojnkpjg, Does the agreement say that they will hand over the proprietary PUID generation stuff as well?

i don't know, you'd have to ask ruaok

or research it

that was what he posted initially

but the track times should be correct if the release has the correct disc id. give or take a few seconds

hmm

when i edit it one at a time the time box is disabled but still white.

so theres no way to apply times to an album with a discID?

move the discid :)

change times, then move it back

ojnkpjg: which redhat office are you out of?

ehh.. i'd have to wait for votes

westford

interesting; i know a bunch of people there, who have no idea who you are :)

times can't be generated from discID?

does it have just one disc id gurt? and how big are the time differences?

yeah one..

that's because nobody would know me as ojnkpjg

i dont know the differences

gotta go LSOT is on

LOST even

who do you know there

caillon, pookstar, ajax, and the formers n and clee

real names?

chris aillon, maureen duffy, adam jackson, dan reed and chris lee

i don't know them

which groups are they in

i've only been there a little more than a month

i think chris and maureen are desktop

adam's "the X guy"

you're cluster & storage?

yep

heh, and "the pork guy" ? :)

yes

cool

wolfmunch is now known as wolfsong13

Hello

hi

Is Nyght there ?

hi

I am pjo, how are you!

My WIndows went totally crashed and now reinstalling

wolfsong13: I haven't seen you in irc lately. I wanted to say thank you fro the support with the automoderator electtion!

Oh, I'm so sorry!

I have never tried moodlogic because MusicIP has done everything I need.

:-) you deserver it

I tried MoodLogic for Jazz.mp3

I can understand your try.

It returned some trumpetter, I remember Clifford

nyght do you think that is #I remember Clifford" ?

I asked Yllona to try and id the jazz snipet. She said she thought it was very similar to Sanborn as well, and she also will be giving it to one or another to see their thoughts as well.

I will look that up and see what I can find. Have you searched amazon to see if there's sound clips there to hear if they're similar

thanks much beth

How do I search for the unknown song in amazon ?