Slashdot Mirror
Using Fractals To Classify Music
Brian McLaughlin writes "A company is working on software that can classify music with fractals and make it easier to find the tunes you want on the Web. Apparently, one can detect the type of music (jazz, heavy-metal, in-between, etc.) by detecting fractal patterns in the music."." I'm looking forward to the day when my music can be indexed and crossreferenced every which way: artist, tempo, year, style, similiarity, heck I wanna know when the Beastie Boys sample The Beatles and be notified and give the option to follow up on the samples within the songs. Someday... I hope.
How can you laud this kind of insult to music? I am utterly insulted by the idea that you could reduce a great art form like music to little numbers and digits. So I suppose we can just forget about all the creativity and emotion that are infused into the works of a musician and classify it as something soulless and robotic like math?
Music may have scientific properties, but it is by no means simply a matter of equations and numbers! Music takes passion, emotion, and raw humanity to exist as such, not a series of bits and bytes! I suppose someone is going try to make a arbitrary music writing program based on this info , no? Well, if so, I'd just like to say that music is NOT math. Unlike math, which takes no feeling or creativity whatsoever (rather robotic computational ability, seeing as how our advanced graphing calculators could easily take the place of any well known mathematician), music requires soul and humanity... something which a computer can never create. A program would never compensate for the genius of Mozart or Bach. Leave it to dronelike programmers to try to rob another great art of its soul and wonder by reducing it to a series of computational processes.
Emerson Willowick: Thinker, Writer, Human Being.
Along this line, you could also tell your player to play other songs with similar fractal signatures to the song you are currently listening to, if you want to hear MORE songs like it - or the exact opposite, play songs which have greatly different fractal signatures, if you want a greatly varying selection of music.
A fellow named Forte promoted a music classification system based on "tone vectors" during the late 70s and early 80s. While useful, in some people's view, for classifying atonal music, the sytem had an interesting side-effect.
All major modes and all minor modes were classified by the same vector. According to Forte, most western music was the same.
So apparently your judging rap, pop, country, classical, etc... from your limited contact with them via Radio? mp3.com? what...? All music has crap in it, but every style has some gems... a few quick examples Underground Hip Hop (anti pop consortium, mos def) Country music (johnny cash, bob dylan) Pop Music (talking heads/david byrne, Pizzicato 5) Classical (Steve Reich, Phillip Glass, Bartok) whatever... open your mind...
I was thinking the same thing.
If you can classify genre and style using fractals, why can't you reverse the process and make a fractal sequencer or drum/rhythm machine?
--The more you know, the less you know.
Mozart did this, actually. Well... not a program, since computers weren't invented yet, but still. What he designed was a set of algorithms -- a dice game -- that one could use to assemble compositions from thematic "primitives".
A number of people have since implemented it in software. I suspect Mozart would be a hacker if he were alive today.
DNA just wants to be free...
........but the Beasties did rip off a Beatles guitar lick for "Egg Raid On Mojo", one of their earliest punk songs. It's available on the original PollyWogStew EP, Some Old Bullsh*t, and the sounds of science compilation. One of my favourite Beastie tunes. I remember freaking out when one of my buddies played me the original Beatles tune......very similar, might as well have been sampled.
"Share your knowledge. It's a way to achieve immortality." -- Dalai Lama
This way Napster can really be a way of finding new artists, try searching for a symphonic rock band that doesn't sound like Metallica, for instance. Or female pop singers that doesn't sound like Brittney Spears.
This could also be a way to avoid misnamed files. Of course, this probably won't be possible until in a few years, when Napster has been sued into oblivion by the Evil Corporations(TM) anyway.
--
Pokéthulhu
Gotta catch you all!
What happens if you feed that program disimilar styles of music, say Metalica, Portishead, and Beatovens 9th?
It fails to find any kind of meaningful consistency, the ATN's knowledge base becomes underpopulated, and the final product is utterly bland and devoid of content. (Wow, so that's how they compose new songs for Britney Spears records!)
By the way... Beatovens? That's a damned cool name for a band! I've got dibs on it!
Also, if I feed the same songs in again do I get the same song out the system a second time, or is it just the same song as the first time?
No, the generative part of the process is randomised, so you merely get a different song in the same style. Look at the example MIDIs in the EMI web pages; there are a handful of generated Nocturnes in there, IIRC, and they're all different.
To the editors: your English is as bad as your Perl. Please go back to grade school.
Stuff like CDDB could be built upon and used to store meta-inofrmation of this kind. Pretty cool if you ask me. XML will be the tool of choice.
I'm looking forward to the day when my music can be indexed and crossreferenced every which way: artist, tempo, year, style, similiarity, heck I wanna know when the Beastie Boys sample The Beatles and be notified
This is a little disturbing to me. Is it really necessary to reduce music to a series of mathematic calculations? Is that really what the aim is? Besides, after that, how far away are we from randomly generated songs (not midi) based on your preferences? That would totally ruin music as an art form, and stifle real creativity. I think this would increase the problem of people just sticking with what they know and not experiencing radically new types of music.
Of course, having hold of this kind of personal preference data is a record exec's dream, so I doubt this company would have any trouble getting funding. Imagine if you could record a song, instantly check it against the personal taste database of 10 million teenagers, and then adjust the song accordingly. What a nightmare.
- Scott
------
Scott Stevenson
Scott Stevenson
Tree House Ideas
I sense much NT in you.
NT leads to bluescreen,
bluescreen leads to downtime,
Such Classification could also be used to detect the type of music you want to hear on a mix. For example:
Let's say you make yourself a bunch of mix CDs and stick them into your 200-CD tray (it makes more sense to have an MP3 player, but hey...), and from those CDs you only want to hear techno, or only want classical because your parents are coming for dinner. The auto-detection functions allow for limitless possibilities for music playback.
Maybe this could also be used on TV to filter out stuff you don't want to see? Like a quick auto-seek for a channel that doesn't have a cxommercial running?!
You might find this bit off of #userfriendly funny in the context of pop music...
I'd like to see this go even further where computers placed around your home can pick up on your mood, and play the appropriate background music. I can definitely see this happening in a few years... vulgrin the MAD
I sig, therefore I am.
You've never listened closely to Bill Evans, then. He played (jazz piano) in a manner that would be nearly impossible even to get on sheet music, let alone reproduce from the sheet music.
Transcribe a solo by one of the great jazz masters. Doesn't matter who. Bill Evans, Monk, Oscar Peterson, Dizzy Gillespie, Coltrane, Dexter Gordon, Sonny Rollins, Sonny Stitt. Doesn't matter. Now feed it into your nearest MIDI program. Play it alongside the original. Gee, doesn't sound so good now, does it? And that's not just because it's not on a real instrument. It's because it's not played by a real musician.
Chess is, by its nature, not a creative act. It's a logic game with a well-defined objective. What's the objective of music? What are the rules?
"The first tuner to use an oscilloscope for display of information was a tube model, the Marantz Model Ten. Beautiful device, tubes. The designer was Dick Sequerra. Later, he started his own firm, his tuners branded under his last name. They displayed the whole spectrum and were popular with radio stations. They are still being made by another company under the same name. Most expensive model about $10k."
I think this is a photo of it, but I'm not sure. Still looking.
Mixing heavy metal and classical isn't really odd. Yngvie Malmsteen put the mainstream spotlight on it for a while in the 80s, and there was also a big revival of neo-classic metal in the late 90s. Happy Helloween-style power metal fused seamlessly with classical on albums like Rhapsody's "Legendary Tales", and darker heavier more traditional metal (think 80s Metallica or Megadeth) combined very well with classical on Rage's "XIII". Then there's the weirder stuff like Therion or Nightwish who have fused metal with opera, or even middle-eastern folk music. I'm just listing the more popular stuff; this is just the tip of the iceberg.
It has even become so common and popular that some diehards are starting to think of it as being "trendy" so there's some backlash against it now. This only got worse when the mainstream alterna-pop band formerly known as Metallica attempted (poorly) to jump on the bandwagon and follow the trend with their S&M album. (Fortunately they didn't know what they were doing so the damage of the movement's credibility was minimal.)
But anyway, metal and classical go very well together. Some people way Wagner (the "Flight of the Valkyries" guy) was a headbanger at heart. ;-)
---
As copyright owner of this comment, I authorize everyone to defeat any technological measure which limits access to it.
It's based on a book published by G Spencer Brown, the mathematical logician (Laws of Form).
It's not quite the same, but the Ultimate Band List won a Webby award for music and has long been accepted as the One True Oracle for music information. Like IMDB (owned by Amazon) it's now more commercial, with "buy this!" links littered all over the place, but it still has decent information.
For more information, click here.
If you can classify music with fractals, and you can generate music with fractals...
:)
Could you fuse categories of music as well?
That is, if you can detect the unique patterns of, say, hip-hop, or classical music, could you feed that back into a program and get some really funky classical music?
This is all pure speculation on my part, mind you. I'd love to help program something like this, but I wouldn't know where to start. I know something about fractals, but very little about music.
(or at least I'm not any good--whatever program I came up with would compose better than I would!)
---
pb Reply or e-mail; don't vaguely moderate.
pb Reply or e-mail; don't vaguely moderate.
Couldn't the reverse, then be done, to take a fractal "fingerprint" of a type of music (say, jazz), add some variables and come up with original music?
It's 10 PM. Do you know if you're un-American?
Like so many announcements, this one is short on details. But music has lots of interesting features which suggest that fractal analysis is a useful start.
Anyone who has browsed through the various Fractal books which actually give you some of the maths, such as the Science of Fractal Images (pub Springer-Verlag) will have noticed the revelation that almost all music has a 1/f^(1-\beta) power spectrum, which is the what fractal approximations to Fractal Brownian Motion tend to head for.
While fractals are supposed to have self similar detail at all levels, and music, digitised or otherwise clearly cannot have 'detail' at all levels for physical or sampling constraint reasons, this doesn't necessarily invalidate the analysis either. The question therefore is what sorts of characteristics are they using as musical indicators. Distinguishing classical from pop is relatively easy even without fractal analysis - the frequency range visited by classical music can be several octaves greater given some reasonable threshold value. The value of \beta may give some insight between styles - estimates for \beta can be as simple as 'distance' travelled by the actual line divided by time, with appropriate normalisation. Anyone care to suggest others? No reason why we shouldn't try and get something working for say CDDB as someone suggested.
Cheers,
Toby Haynes
Anything I post is strictly my own thoughts and doesn't necessarily have anything to do with the opinions of IBM.
When he finished, he realized that the comments were much, much longer than the code they attempted to describe.
;) Seriously, even if I expand the above formula to its textual correspondent, it'd still be much shorter than it could be expressed in any algorithmic language, because these are languages for computation, not for mathematical abstraction.
;)
(Oddly enough, I just had a similar discussion in DALNet #perl. I'll be a good programmer, then, and practice code (example) reuse.)
I remember that post on PM too; I disprove it thus.
"This is left as an exercise to the reader": write a Perl module Math::MB, such that, after
use Math::MB;
tie($m, 'Math::MB');
$t contains the Mandelbrot set. The length of Math::MB must be strictly smaller than the following description of the Mandelbrot set:
Definition. Let Z_1_(c) = 1, Z_n_(c) = Z_n-1_(c)^2 + c; then M = { c e |C | lim_n->oo_ Z_n_(c) * oo }.
(_foo_ denotates a subscript; oo denotates infinity.)
(Thus it is proved that, however concise Perl may be, mathematics is even more so.
My point is that the same thing applies to poetry. It shouldn't be described in terms of the pure textual size of its written form, but in terms of the "size" of the symbolic structure that the reader gathers from reading it. Perl can represent, e.g., pattern-matching, array looping and I/O, concisely because that's all that a Perl program does; there's no additional level of significance to the "text" that is Perl code, so the issue is merely one of Perl syntax vs. English syntax. However, ideas that are easily expressed in a "real" language like English, to a human, such as the concept of limits, the concept of a set, and the concept of a logic variable, have no reasonable equivalent in Perl programming.
Thus, Perl may be beautiful, but it's no poetry.
To the editors: your English is as bad as your Perl. Please go back to grade school.
Actually, this could be used to shoot down the popular "filename says sandman, but is not metallica" arguments. When Metallica told Napster to ban 300K users, it was because of a filename which does not infringe on copyright. Now a band can say, "Hey, not only does it have a suspicious filename, but it follows the pattern of our hit song exactly.
This is cool technology though...maybe someday we will be able to prove for certain that Vanilla Ice did use David Bowie and Queen's 'Under Pressure' in 'Ice Ice Baby'. Hmmm
They've changed their name to Gracenote. This is mainly just a way to relaunch their brand since they now work on more than CD's - hence the old name is too restrictive...
It's actually not that difficult. The algorithim (that I know) is a fairly straightforward neural network. I took an Intro to AI class while I was in college - had a friend who's final project was a system that could determine the type of music on a CD that was currently in his CDROM drive.
He trained it first, with 2 CDs from each genre from his collection. He then went through and had the system analyze and 'guess' the type of music. From his collection of ~100 CDs, it correctly identified the different types about 95% of the time.
Now, I don't know the constraints that these folks are putting on their software. But if Erik could code up a working model for a 200 level AI class, I'd hope that this company can handle details.
JJ
My favourite music is Einstuerzende Neubauten, and anything else just isn't noise.
Any music which you are not familiar with will sound "all the same" until you are familiar with it. People have said that about jazz, pop, classical, heavy metal, techno, hip hop, punk - anything in existence. If you think any genre "sounds all the same", I submit that you are not familiar enough with it to be able to understand what the different artists are trying to achieve. I would say that you need to be familiar with at least 100 recordings of a genre, and have read several books about it, before you are qualified to even begin considering judging it.
...) are no different.
Something like country is tremendously diverse, and is also one of the oldest recorded musics. Few music lovers don't love older country to begin with, and when you add things like alt.country and bluegrass and contry-folk (each of which have a bunch of different sub-genres), you have a tremedously respectable music, and inarguarbly one of American's two or three finest traditions.
Of course, if all you've heard is Shania Twain and whatever else they play on the radio, you've missed out. Like any genre, the best country music is not played on the radio. Judging country music by Garth Brooks makes about as much sense as judging jazz by hearing only Kenny G, metal by only Bon Jovi, rap by only Snoop Dogg, and classical only by Charlotte Church. For any genre you need to dig deeper than the tunes played on the radio, and country (and hip-hop, and jazz, and
Based on what the article says, I'd expect Torn and Eno to be lumped in with Yanni and Tesh. The "amplitude fractals"--or whatever they're calling them--would be about the same.
But, if they combined the amplitude analysis with something that could determine harmonic relationships between simultaneously occurring tones, they might spot the difference (Eno and Torn being "uglier" to most people's ears, because they use more complex, close-on-the-scale voicings, generally). But it still wouldn't know the difference between the new Deftones album and the second Sunny Day Real Estate record, based on that.
So, if you can hear the difference between, say, John Fahey and Roy Clark, something like this would be useless to you. Probably a near-"true AI" problem, if they were bothering to get it right (like, identifying the guys in Birdsongs as members of Mission of Burma based on their styles). Which they aren't. They're just whoring for VCs by using obfuscatory tech-talk to hide the uselessness of their allegedly existing product.
Your mouth is like Columbus Day.
I was skeptical but it actually works quite well. Even better, they're an open source shop through and through: Apache, mod_perl, mysql, etc.
Advertisers, psychologists, neurologists, and pharmacists have all learned ways to manipulate the brain in very mechanical ways, getting very machine-line responses.
It's romantic to think we each have a "soul" and I wish I could believe it. But as science progresses, the concept of the soul starts to look more an more like just another archaic model for explaining processes that we used to not understand.
Music is going to be demystified some day, and the torch will be passed from the artists who do it intuitively, to engineers who do it very consciously. Then they will write computer programs to generate it and machines that are made out of meat will listen to it and smile predictably.
Sometimes I hope that by the time it happens, I will be dust. It seems as if understanding humans will dehumanize them. In a few hundred years, whenever someone mentions the old terms "dignity" or "free will", people will chuckle and say "those are relics of an outmoded belief system."
---
As copyright owner of this comment, I authorize everyone to defeat any technological measure which limits access to it.
it definately won't be able to recognize good music over bad. But a (broad) categorization should be possible:
- rock: includes punk, hardrock, speedmetal
- techno: includes trance, gabber
etc.
Obviously a lot of people are going to be unhappy, getting 'their' music lumped up in 1 category. But it could still be a useful tool.
//rdj
No one can understand the truth until he drinks of coffee's frothy goodness.
--Sheikh Abd-Al-Kadir, 1587
Your favorite type of music is the only real music around, and everything else is just noise.
luckman
luckman
I don't involve myself with flames, much less know how to bait one.
Sure you can. You can define something as intensely personal as feelings through somethign as crude and impersonal as words. They'll be able to write an algorithm that will contain what's different about X piece of music from Y piece of music. Do that for a while, and do it well, and you'll have a lot of useful and interesting data to correlate.
You can reduce a piece of music to a generic set of characteristics by writing it out in SHEET MUSIC. It isn't that hard. I doubt that an algorithm would always categorize things the same way as you would, but I bet it could still do a good job, or come up with some insights.
For any further dissenters: the "humans are intrinsically superior because we are human" crowd said the same thing about chess. Where are they now? Oh, they just jumped to a new topic.
Jeez. If you're not going to contribute, but just want to sit on the sidelines and talk about how people shouldn't be able to do WHAT THEY ARE DOING, shut the hell up and go somewhere else.
---
pb Reply or e-mail; don't vaguely moderate.
pb Reply or e-mail; don't vaguely moderate.
i suppose that rhythm would carry more weight than harmony in that kind of analysis... like straight ahead jazz can be recognized by swing eith notes, whereas fusion wouldn't swing but would still be modal. or, what is the difference between a three chord ac/dc and alot of the punk stuff... probably tempo and tightness. interesting to ponder....
I think you're afraid that this might work, that science might invade the realm of art. What you don't realise is that science and art are two ways of expressing the same truth.
If a mathematical mechanism can produce art, it doesn't mean that art is fundamentally ugly. It means that mathematics is fundamentally beautiful.
----
Using fractals to find simalarities in music should allow us to prove that there is only one rap song and that it is just being re-recorded by everyone with a microphone. Same for most classical, country and pop.
I recall there being a high-end, all-analog radio tuner that used a special wide-screen, green phospor, cathode-ray tube to display the entire FM radio spectrum at once. The nice part is, it allowed you to tune to the center of what was being broadcast. (Those big transmitters did drift.) And it totally fit with the analog-only mindset of being forced to listen to a digitally/decimally perfect frequency.
The long-time users of those systems said they could tell what type of music the station was playing by the frequency distribution, and frequency energies being used. Some said, for their favorite station, they could even tell what period of music was being played, or if it was one of their favorite composers.
there are places where tb is more common than tv.
If you've seen interviews with many artists, you've probably noticed how most of them hate to be classified. If this really works, it would kill a lot of pretension. I can see it now:
Reporter: Despite the fact that you're considered a rock artist, you seem to be having a great deal of success amongst the country and even R+B fans. Why do you think that is?
Artist: Well I don't really think of myself as a rcok musician. People are always trying to classify my music as pop, or hard rock, or soft rock, or whatever. But I don't restrict myself to those terms, I just think of myself as an artist, and I think my work really defies being simply classified as rock.
[Reporter looks down at a laptop, hits a few keys]
Reporter: No, the computer says it's clearly rock music. Not country, not big-band, not funk, not innovative-genre-transcending-art, just rock music.
Artist: Well, yes, but it's really-
Reporter: That's all the time we have for now. Tune in tomorrow as I interview another popular rock band.
Would be using this software to analyse what I currently listen to when i'm doing certain tasks on my PC. Then it could download/compose/stream/remix music on the fly to suit my tastes all day long.
:)
No longer would I sit at napster wondering which (unsigned) artist I should download today
I hope it's better than the BAIR system for recognising naughty pictures (http://slashdot.org/article.pl?sid=00/07/21/1516
</skeptical>
Mike.
Tales from behind the Lagom Curtain
That is the shortest story I think I have ever seen on slashdot!
I can't myself see how much detail can be garnered from the amplitude alone of the notes in a musical piece. I can see how heavy metal and jazz would be quite different, but how about jazz and drum'n'bass, they are incredibly similar forms that would be distiguishable by the underlying beat rythms (or perhaps more likely through the persistance of instruments). Would a fractal based on note amplitude grab this? I can imagine it might, but if you threw it the whole gamut of dance music (acid, trance, acid-trance, garage, girly garage, ambient, techno.....need I continue) I am sure it would fall apart as the large scale use of compression alone would bring these musical forms incredibly close together in terms of amplitude (if everything is at 100db post-compression as often happens with dance music).
Another useful bow in the arrow of anyone interested in categorising music, but I feel that a full quiver of tools is always going to be needed to even come close to trying to do this job.
Never underestimate the dark side of the Source
This could create one hell of a visualization plugin for Winamp...
Sleep: A completely inadequate substitute for caffeine.
There is a strong correlation between average spectral usage and genre, particularly between genres such as classical (which looks fairly symmetric, as a good symphony uses the whole spectral range) and more beat/vocal heavy types such as electronic, which are dominated by low frequencies. For song level identification, however, you need to include additonal features, such as sonic energy, which gives a unique (across my test sets at least) fingerprint of music. But, based on those sonic signatures you can realise that track1.mp3 is really Metallica's Unforgiven (which is useful if you didn't know that it was ;).
This reminds me of a post I saw on Perl Monks. The author described how, in one of his English/Literature classes, his teacher defined poetry as being a succint and condensed use of language. Later, while writing perl, he was commenting everything and trying to describe the script completely. When he finished, he realized that the comments were much, much longer than the code they attempted to describe. Therefore, he said, perl (arguably this could be extrapolated to programming in general) is poetry.
heck I wanna know when the Beastie Boys sample The Beatles and be notified and give the option to follow up on the samples within the songs.
As someone who routinely beats samples into a whir of unrecognizable sound, and would prefer not to pay any "sample clearance fees" whenver I sample anything. You want a rights issue, TacoBoy? You are NOT free to sample anything you want to. Maybe you might want to read what negativland has to say about Intellectual Property.
Bad things often happen to good people,
It is up to them to see that they remain good.
I'm currently listening to "Pulse Piece", by Birdsongs of the Mesozoic. Layers of distorted piano, organ, and percussion, with a pulse-like beat. I wonder what this clever technology would make of that? Rock? Jazz? Techno? (It predates techno by a good ten years)
Hell, i wonder what it would make of half the stuff i have ripped to mp3 here at work... Steve Tibbetts, David Torn, Erik Satie, Brian Eno, Jean-Michel Jarre, Klezmatics, Last Exit, Nordic stuff, Sonic Youth, etc...
--
Hand me that airplane glue and I'll tell you another story.
Wouldn't it be nice if Slashdot could use this technology to eliminate the overload of spam and trolls? If it can work for music, it can work for spam. Just my 2 cents as an honest AC. [Don't blame ACs for spam, now the fashion is for spammers to take a nom-de-fume -- I won't dignify them by mentioning their names.]