why is there no identifying fingerprint feature that can be enabled/disabled in limewire for audio files? it seems this would open up a multitude of possibilities to users of limewire exchanging files over gnutella. we wouldn't have to rely on metadata such as file names or ID3 tags to identify files, and we could also resume downloads from other users that have a matching fingerprint. there are also many applications for music discovery that could come out of implementing such a future.

what's the status of the fingerprinting scene anyway? I know MusicBrainz was maintaining an open source database of them, many of the other fingerprinting companies went down with the dot bomb crash. relatable partnered with napster a while ago, and I think is still going. cantametrix and etantrum went out of business, maybe there are some assets left over limewire could license with proceeds from sales of limewire pro?

anyone know? any ideas? am i smoking crack?

-justin

Don't do accustic fingerprinting. Just use md5 checksums or some other hashing algorithm. It would work better, and not be exclusive to audio files.

Great concept... just needs some generalization + refinements...
Here's my offering... brainstorming... I'd love to beta test any prototypes!!! (I'd rather write specs and do the testing, rather than the coding... )

Shareaza is already doing this, please see:

http://www.gnutellaforums.com/showth...224#post133224

__________________
Geo here... <small>antique computer enthusiast
Mac Powerbook G3/500MHz=<i>Pismo</i> :: 384MB ram; 80GB hdd; <b>MacOS 9.2.2</b>
Compaq Presario 5360 :: 450MHz AMD Athlon, 192MB ram; 120GB hdd; <b>Win2k PRO SP4</b>
ISP :: <u><b>SBC</b>-Yahoo!-DSL</u>; 164KB/sec DL; 50KB/sec UL; <u><b>2Wire</b> <i>DSL modem / Wi-Fi / NAT firewall</i></u>
PtoP :: LimeWire 4.8.1 <b>PRO</b> || Shareaza 2.1 <i>for BitTorrents, mostly</i> || but not concurrently! <i>
This is an INTEL-free, pre-Y2K CPU, recycled-computer environment! </i>

I'm not sure about this concept... The way LW displays results, only the first result will diplay the metadata to the user, the user will have to expand the list of one result to find the metadata he *thinks* is the best, because he haven't dowwnloaded it yet.... I think it is too complicated for the average user (believe me on this). The best way is to download the file with the name that fits you the best even if many results are in fact the same mp3, evaluate if it is the good thing by downloading it and THEN change the metadata if you think it doesn't suit the file. You know what, a rating system of the quality of files like the one implemented in credence http://www.cs.cornell.edu/People/egs/credence/ is something we want much more. Another problem is the hasing of the file in two parts with and without the metadata, this becomes complicated without advantages if you ask me.

You can see the arguments used for and against at the GDF when seraching for "mp3 strip metadata", http://groups.yahoo.com/group/the_gdf/

Ciao

__________________
Liens d'intérêt /Links of interest:
Gnutellaforums en français /The House's rules you have to respect / First search the forum, then create a thread / Free software alternatives! - Logiciels alternatifs gratuits!/