Picking up on the PMA side of things, just how does it decide which tracks to recommed? Has anybody else used this and do they have any comments on its accuracy/taste :-)

Gavin

Are you still working on this?

Sean

Not actively. The service is still running and will probably do so for the time being... but my company went the way of the dot com do do, and a majority of employees (including the entire dev team) have been laid off.

Any updates to it are purely done for the fun of it now. :)

(O|||||O)

Im sorry to hear that you got laid off. Who has the database, and will it keep running? Im trying to get the author of Dr. Tag to include the ability to ID songs using songprint individually. Do you think something like that would be possible?

Sean

Yeah, I was talking to the Tag Studio guy about it too. he said he was interested but probably will not be as much, with curent events being what they are.

DiGNAN
Ob-La-Di, Ob-La-Da, etc.

Well, keep in mind, the technology is VERY stable, and we still are looking for partnerships/licensing options. I still am working on the project, helping a few of the old employees maintain the servers and so on. It just isn't as high of a priority as it was 6 months ago.

We have more than quintupled the size of our database in the past few months, and we still continue to have new fingerprints added daily.

(O|||||O)

Err, ummm, how to put it tactfully... Does your company, should it reach the, ehm, end of its lifespan, plan to opensource the technology and put collected database in public domain, or do you, hopefully, consider such discussions premature?

Anyway, good luck!



Dragi "Bonzi" Raos
Zagreb, Croatia
Q#5196, MkII#80000376, 18GB green

I would appreciate it if you could email me a compliled win32 exe. I still havent figured out how to compile the thing. I guess im not getting the environment settings right in visual c++.

Sean

They are very premature. We still have several companies interested in acquiring the technology and/or licensing it... but these kind of contracts take forever to go through.

If anything along those lines happens, I'll let ya'll know.

(O|||||O)

I put together an ugly little shell script that will call songprint and add the tag to the MP3. I tested it in Cygwin, but it should work in any unix environment. It looks to see if you have an id3v1 tag and if so, preserves the year (because SongPrint doesn't provide that info). If SongPrint returns multiple albums, it uses the one that was already there, if it was in the list SongPrint returned, otherwise it uses the first one returned. That may have been confusing, sorry. It doesn't overwrite the old tag, just adds another one after it. It is exteremely slow due to the fact that accessing the SongPrint database is extremely slow (but low CPU utilization). It took me about 8 hours to id 2858 songs, of which 1214 were recognised. Thats about 42%, a little lower than I was expecting. And my music is not all that weird. ClemsonJeep said in September of last year that there were 60,000 songs in the database and growing daily, so maybe others will have a higher percentage than I did. You can periodically run tagall to tag any new mp3s you have aquired and/or attempt to tag any mp3s that were previously unrecognized. It will not attempt to re-tag mp3s already tagged by this tool. This will make subsiquent calls to 'tagall' faster, depending on how many songs it recognized the first time around. It uses the comment field to tell if it was already succesful for a given mp3.

And here is 'tagall'

Cool. It's wonderful to see that people still find our tool useful... We had a new version of the library in development that was 200% better than the one that is in production now, but alas, will probably never see the light of day.

You don't happen to be (or know any) a VC with deep pockets? ;-)

(O|||||O)

cool scripts. Too bad the only linux platform I have is the empeg. I have located another program, freeamp, that uses technology developed by relatable to id the songs.

Sean

Whats a VC? Negative on the deep pockets.

Terminator, thanks for the tip, I'll check out freeamp.

VC=Venture Capital, companies whose sole business is investing in other companies.

___________
Tony Fabris

Venture Capitalist - they are non existant these days. The library freeamp is using is pretty big. You can get more info at http://www.musicbrainz.org
I just checked their server stats - they have over 250,000 songs in their library,
Sean

Edited by Terminator on 26/03/01 06:58 PM.

So, ClemsonJeep, is there any songs currently being added to the database? Why not do The Right Thing and put the database and server-side software in the public domain and let this cool technology flourish? You guys had your chance to commerialize it. It didn't work out. If you continue to keep it secret, some other company will just write there own version of it. It sounds like the FreeAmp people already have. But I don't know anything about FreeAmp, only what Terminator was saying.

Anyone who loves Linux but mostly runs Windoze for multiple reasons (like me), needs to check out Cygwin. Its basically a bash prompt for Windows. Complete with all the great shell tools normally availible at a unix prompt that only Bill and God know why were never included in any version of Windows.

Okay, I just looked at the FreeAmp, Relatable and MusicBrainz websites. The quiestion I have that I wasn't able find the answer for on any of the websites is, Is there a tool that can leverage this large database to fix your ID3 tags for you? If not, is there a tool that will just output the correct information, like SongPrint?

Run free amp, hit the my music button and I think free amp will search the whole computer for mp3s. When its done, go to the relatable tab and create a profile. Then hit start signituring. It will look up everything, and I THINK it tags everything that doesnt have an id3 tag with a ID3v1 tag. If you go to the musicbrainz page, you can download the source for the client.

Sean

Edited by Terminator on 26/03/01 07:25 PM.

Cygwin is really cool! It looks and works just like linux. Maybe I will be able to try out your shell scripts after all.

Sean

Yeah, Relatable was "competition" for our company when we were around last year. I'm not sure if they are profitable or not, but at least they are still around. :)

(O|||||O)

Well, we are still in the process of talking to a couple companies about selling the IP of our company, so the technology can still evolve. If those bomb out, I'll be the first one to post the code for everything to sourceforge. :)

And yes, the database still evolves. It was designed to work with little-to-no human interaction. If you tag a song that isn't in our database, it attempts to figure out information from ID3 and the filename itself, justifies it all, then sends the data to our server which goes through a process we call "alignment" which makes sure we get little to no duplicate or bad data. Thus the database is still growing.

(O|||||O)