Monopolising music recommendation
p2pnet news view DRM | P2P:- “There’s a company called Music Intelligence Solutions (MIS) (formerly Polyphonic HMI) that’s attempting to monopolize machine-based music recommendation and measuring,” says a friend of mine in an email.
Describing what amounts to a kind of DRM, “This type of technology is used to analyze the acoustic properties of a song and then recommend other songs that sound similar,” he says, pointing out the news should be both interesting and relevant for anyone who’s into this kind of research, or who has a business depending on their ability to, “not be beholden to potentially onerous intellectual property”.
He goes on »»»
It’s also used in what’s commonly referred to as fingerprinting, which is what the popular iPhone app by Shazam does when it uniquely identifies a song you’re listening to.
A first patent was granted to MIS in 2006 and a second that would further cement MIS’ stealthy grip is currently at the USPTO pending approval.
There seems to be a great deal of prior art that’s being ignored by the USPTO, or that simply hasn’t emerged. If it isn’t called into public view and clearly posted where the USPTO reviewers can easily find it, there’s a serious risk that much of the research in the field of music information retrieval will be gated off by this company.
Researchers and others working in this field need to be made aware of this so it can be addressed. Putting forth all the prior art (PRIOR TO MID 2001) seems to be the key to blocking this intellectual property grab.
And to even overturning the already existing patent, which likely should never have been granted in the first place, if a thorough review of prior art had been conducted or brought to the attention of the USPTO reviewers.
It appears there are a number of companies and investors who should be interested in taking a close look at this including Music IP, Fraunhofer Institute, Double V3, BMAT, Echo Nest, Queen Mary University, and plenty of others.
The patent in question can be found here: http://www.google.com/patents?id=N0R6AAAAEBAJ&dq=7081579
Let’s take a close look at Claim One…
Claim Number One in the patent titled “Method and system for music recommendation – Vicenç Gaitan Alcalde et al” seems to attempt to exclusively protect and would prevent anyone from doing any type of machine-based music recommending. If you’re extracting and processing ALL OF THE SAME features MIS is extracting, regardless of whether or not you’re extracting additional features, you would seem to be in violation of Claim One. If you’re employing steps A through G (see claim below), you may be in violation of Claim One. Readers of this post should confirm this: It’s not enough to be doing more (a superset) than what the claim herein describes. You must be doing something different! If you’re doing ALL of the things (inclusive) described in this claim, you may be in violation.
This is the Actual Claim from the MIS patent with a few added notes in brackets for clarity
A method of analyzing music, said method comprising the steps of:
a) providing a digital database comprising a plurality of digital song files;
b) selecting one of said song files for analysis;
c) dividing said selected song file into a plurality of discrete parts;
d) using Fast Fourier Transform techniques on each part of said selected song file to establish a plurality of coefficients, wherein said coefficients are representative of predetermined quantifiable characteristics of said selected song, and; wherein each said predetermined characteristic is a physical parameter based on human perception including:
brightness;
[Brightness is quantified as a spectral histogram built from the FFT.]
bandwidth;
[Bandwidth is quantified as the variance of the spectral histogram built from the FFT]
tempo;
[Tempo is quantified as a measure obtained using a "Beat Tracker" algorithm (based upon autocorrelation values). The Beat tracker algorithm calculates how many beats per second are in a "chunk."]
volume;
[Volume is quantified as the average change in the bit sample amplitudes at lag 1. Lag 1 refers to consecutive beats.]
rhythm;
[Not quantified]
low frequency;
[Low Frequency is quantified as spectral power weighted with ten inverse frequency.]
noise; and
Noise is quantified as the sum of all the FFT coefficients normalized to the maximum FFT coefficient.
octave, and [Octave is quantified as the temporal mean of Cepstrum (inverse of the logarithmic FFT). The Octave is calculated for each chunk and averaged.]
how said parameters change over time;
e) determining an average value of the coefficients for each characteristic from each said part of said selected song file;
f) compiling a song vector comprising a sequential list of said average values of the coefficients for each said characteristic for said selected song file; and
g) repeating steps b) through f) for each song in said database.
And it gets more severe, says the email, adding »»»
In a subsequent patent (pending) filing (link below), MIS has tightened the language to prevent anyone from using even a single feature analyzed and extracted from a music file.
Take a close look at the last sentence. The word “including” has been changed to “consisting”.
http://www.google.com/patents?id=mVCZAAAAEBAJ&dq=11/492,355
Original Clause D
d) using Fast Fourier Transform techniques on each part of said selected song file to establish a plurality of coefficients, wherein said coefficients are representative of predetermined quantifiable characteristics of said selected song, and; wherein each said predetermined characteristic is a physical parameter based on human perception including:
New Clause D
d) using Fast Fourier Transform techniques on each part of said selected song file to establish a plurality of coefficients, wherein said coefficients are representative of predetermined quantifiable characteristics of said selected song, and; wherein each said predetermined characteristic is a physical parameter based on human perception selected from the group consisting of:
Researchers and companies wanting to preserve their rights to use and capitalize on these music analysis techniques for use in recommendation, “only need to reveal and draw attention to prior art (prior to mid 2001) that demonstrates that CLAIM ONE (exactly) was practiced in the field prior to 2001,” says my mate.
Stay tuned.
April, 2009
Use free p2pnet newsfeeds for your site. It’s really easy! Subscribe to p2pnet.net | | rss feed: http://p2pnet.net/p2p.rss | | Mobile – http://p2pnet.net/index-wml.php
Net access blocked by government restrictions? Use Psiphon from the Citizen Lab at the University of Toronto. Go here for details.
May 1st, 2009 at 1:40 pm
http://isen.com/blog/2009/04/broadband-without-internet-ain-worth.html
Interesting post about broadband, Internet and Net Neutrality. Via dslreports.
May 2nd, 2009 at 11:25 am
MusicIP was doing fairly similar stuff at least as early as December 1999.
May 2nd, 2009 at 1:05 pm
I will create a device that will analyze the sound waves travelling the atmosphere on earth the exact same way human ears do and patent this and then anyone being caugh listening or hearing anything will be infringing on my IP; I will suit them!
Get a grip; you cannot put a patent on what a process does, but only on how it does it.
May 2nd, 2009 at 4:41 pm
I used almost exactly that same feature set for genre classification in my master’s thesis, published on my site in 2000 http://www.sethoscope.net/aimsc/ . I have zero signal processing background, so it’d be fair to say those features would have been obvious to an actual practitioner back then. George Tzanetakis was doing similar (and more) things with Marysas at the same time.
May 3rd, 2009 at 11:43 am
Wendell, can you substantiate that or get someone from MusicIP to post publicly available evidence of that? My understanding is that the substantiation has to have been made public before mid 2001 for it to be valid.
Seth, it looks like you did not publics your work until 2002. Am I reading that wrong?
May 3rd, 2009 at 11:58 am
Captain Sparrow – You’re not addressing the issue. The patent in question has already been granted. Do you have specific information you can put forth that will / would overturn the grated patent in this field?