Audible Magic blocks P2P song swapping...

[Roundabout]

From ZD net:

Complete article here.

From Napster's death to Audible Magic

The idea of filtering file-swapping networks got its first test run in Napster's last days, when courts mandated that the company block trades of copyrighted songs with near-perfect accuracy. The company first tried to block key words, but that failed when users simply renamed their songs.


Later, it began blocking using audio "fingerprinting" technology supplied by partner Relatable, and the amount of material available through the service dropped from tens of millions of files to just a handful almost overnight. Napster closed its doors to the public not long afterwards.


Audible Magic's song-identifying technology is the product of a group of former Yamaha sound engineers, who originally created the software to help movie post-production studios search massive databases of sound effects such as footsteps or door slams. In the late 90s, they joined forces with former Hewlett-Packard marketer Vance Ikezoye and his newly formed Audible Magic startup, and turned their attention to identifying digital media files such as songs.


The company's technology works by identifying "psycho-acoustical" properties -- essentially the computer equivalent of listening to the song itself. That means that the identification procedure is flexible. A song might be compressed into a lower quality recording, or have a few seconds of silence taken out at the beginning or end, or be otherwise transformed, and the technology will still recognise it as the same song, the company says.


The identification technology has already won credibility, through being used by songwriters' and publishers' trade association SESAC to identify when songs are played on broadcast radio in order to collect royalties. Several CD pressing plants also use the technology to track what they're manufacturing and ensure that their customers aren't trying to create counterfeit discs.


But it has been the company's peer-to-peer-focused efforts that have now brought it squarely to the forefront of the copyright debates. Audible Magic is offering two different versions of its technology, one focused on networks and one on file-swapping software itself. For several years it has tested a network-based "appliance," which would sit inside an Internet service provider (ISP) or business network and monitor data traffic as it goes by. If it identifies a copyrighted song, the technology would stop the transfer in progress.


A test of that technology was held at the University of Wyoming last year, but was ended after students complained about privacy invasions. In response, Ikezoye offered a university-focused version that simply blocks the copyrighted songs, and does not link specific trades to specific computer users. That's helped spur new interest in the technology, such as from the University of Rochester's Phelps, although announced customers are still few and far between.

Inside your software?

The company's main demonstration for the past few weeks has been a version built into a piece of open-source Gnutella software. Similarly, it could be built into any other popular file-swapping package, company chief executive Ikezoye said.


In that software-based version, the technology watches what songs are being downloaded, and when it has enough data to make a match -- usually about a third to half of the file -- it uses the Net connection to call Audible Magic's database. If it finds a match with a copyrighted song, it stops the download midstream. Similarly, when files are put into a shared folder, the demonstration software calls up the Audible Magic database. If it finds a match, it prevents the song from being shared with other people on the network.


That second version of the software has not been tested on a large scale. While it appeared to function well in a single-user demonstration, implementing it on a widespread basis, particularly in software such as Kazaa or Morpheus where tens of millions of search requests a day are made, could have unforeseen consequences. Moreover, for the filtering to work on a large scale, Ikezoye said that pressure -- probably through legislation -- would have to be put on file-swapping companies, which would be unlikely to voluntarily adopt his technology universally.


"This implementation clearly requires the cooperation one way or another of the peer-to-peer vendors," Ikezoye said. Audible Magic's technology is far from perfect, even if it works as demonstrated. It's most critical weakness is likely to be encrypted files and encrypted networks, which its audio recognition software can't break through. Nor is it difficult to imagine hackers creating "cracked" versions of file-swapping software that have the song-recognition technology broken or stripped out, if legislators were to mandate its use.


Audible Magic is not the only company seeking to build filters for file swapping. Napster creator Shawn Fanning's new company Snocap is working on similar technology, with an aim of giving record companies and music studios a way to make money from peer-to-peer networks.


But the file-swapping controversies are today as much rhetoric and politics as they are technology, and the last few weeks may have quietly seen a change in the file-swapping debates.


"I've achieved my objective, which is to say our technology works," Ikezoye said. "It is interesting that the question has shifted from 'is this possible?' to 'how should this be deployed?'"

[BJ_M]

record it backwards .... :lol: :lol:


just kidding -- i have no idea ....

interisting technology ....... what happens when someone blocks the server ?

[The village idiot]

Simple.... Use something like winrar to break it into smaller pieces. would work best if used on full length CD's as opposed to single songs. One more step is to use software to generate parity files (smart par, quick par, quick par2, FSRaid) of that archive. Just make the recovery to 100%, and then never send the original files, just the parity files. Then recover the parity files into the original archive.

In the first way, the blocking software would have to be able to scan within archives, not a difficult task.

In the second, since each par file has bits of info about each portion of the archive (chunks), you couldn't just scan the chunks to tell which song is which.

For a twist on the first part, after creating an archive with winrar, you make a parity set. Then you rename the archive chunks with random file names for each chunk. Winrar will not be able to put the chunks back together, so scanning inside the archive will not work. Then you use the parity set to fix the filenames, and then extract the archive. You could also use passwords on the archive and/or encryption, but that is a pain in the @$$.

Once again, technology broken before it is even released!

Pretty neat software though... but they really should just stop trying. The more difficult they make the process, the less controll they have over the way things are swapped.


I have an idea...



.



.



.



.


.



.



.




.




.


.


.



.


Just shut down the internet!

[housepig]

screw it, I'll go back to sneaker net.

you can't block me "uploading" a disc of mp3 files to my friend's hand...

[Gazorgan]

Simply do on-the-fly encryption between peers. Unique random keys for every chunk. Very simple thing to do.

Send the file in random order, then sequentially assemble it.

Send 'fake' blocks at some determined interval. This increases the size but it breaks the song up a lot.


The list goes on. I'm sure there are some royalty rights in there somewhere......

(and how can you scan 4,675,235 GB of Kazaa files?)

(and who is going to pay for the TB's of bandwidth needed daily to do the checks?)

[dphirschler]

I guess I should start buying my music now... HAHAHA. Just kidding.

**** them. I am through with music. Hell, I can't even whistle now days without the RIAA breathing down my neck wanting royalty payments! I don't even like listening to the radio anymore. As if the frequent commercials weren't bad enough, they make music no fun anymore. Listening to music now makes you feel guilty. They can eat me.


Darryl

[zworg2]

they make music no fun anymore. l

Yup, but then most of the stuff released these days is as much music as a big pile of luminous rice is art!! Virtually all of the music (real stuff, ie more than 5 years old) that i want to listen to has already been downloaded or purchased.

SCREW EM

(and who is going to pay for the TB's of bandwidth needed daily to do the checks?)

The end user, those that download will probably have theri ability to download reduced to encompas this extra work being done by the software.!

SCREW EM

Simply do on-the-fly encryption between peers. Unique random keys for every chunk. Very simple thing to do.
Send the file in random order, then sequentially assemble it.
Send 'fake' blocks at some determined interval. This increases the size but it breaks the song up a lot.

Yup anything they can do we can do better.....

SCREW EM

THREE STRIKES ---- THIER OUT!!

[sullen]

What about SHN and FLAC compression? Will this program identify these compressions since you have to have a plug-in to play them compressed? What about Bit Torrent? That type of download is almost always randomized until the full set of files is downloaded. I think that this will have some positive effects though in that it will better seperate legitimate live recordings from copyrighted songs allowing live recording traders to flourish.