h a l f b a k e r y"Look on my works, ye Mighty, and despair!"
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
browse anonymously,
or get an account
and write.
register,
|
|
|
What this is is an audio renderer of a simple but as yet undiscovered fractal called The-Set-of-All-Music. Every song that has been or ever will be composed can be played just by tuning the device, which may have a CPU but look like a radio, to the 4 coordinates of the song in music-space. ... I know
there is fractal music out there but it very much disappoints me because it was generated in MIDI, and I am talking about something that mathematically generates a waveform, as the fourier series does (an example of an Excessively Complex formula that actually can generate ALL MUSIC!). --- This also is useful for making a smart piano that makes music when you push any keys you choose that light up, which indicate that the notes that are lighting up are part of the music set and therefore will sound good when played after the last notes you played... so any dummie can play the piano.
fractal being played
http://www.ultravires.net/notnoise.mp3
discovered in a noise generator when I was 12 [mr2560, Oct 04 2004, last modified Oct 05 2004]
Mozart composing with Dice
http://sunsite.univie.ac.at/Mozart/dice/
Not exactly what I'm talking about but interesting [mr2560, Oct 04 2004, last modified Oct 05 2004]
Eigenradio
http://eigenradio.media.mit.edu/
A good idea... but it sounds awful! [mr2560, Oct 04 2004, last modified Oct 05 2004]
Fractal Music Gallery
http://thinks.com/sounds/fractal.htm
[DeathNinja, Oct 04 2004, last modified Oct 05 2004]
fractal being played (cut for play as a loop)
http://www.ultravires.net/revent1d.wav
also from SMSP/noise generator, richer sound [mr2560, Oct 04 2004, last modified Oct 05 2004]
A Lyrical phenomenon
http://dictionaraok...me_with_science.mp3
[mr2560, Oct 04 2004, last modified Oct 05 2004]
A monkey with a calculator (is it sound?)
http://www.ultravires.net/POCKET.WAV
It's really just a text file with a number in it. [mr2560, Jan 10 2005]
Please log in.
If you're not logged in,
you can see what this page
looks like, but you will
not be able to add anything.
Annotation:
|
| |
Bad science. No science at all, in fact... just magic. |
|
| |
Nope. This "magic" Works! First, when I was 12 , I accidentally discovered a "set-of-some-music". An old recording of one of it's compositions - which came directly from the device to a speaker - can be heard at the link. This set is a known fractal today. Second, I have determined that every song can be coded into a number... usually a very large one but not necessarily. (A nearly infinite number of numbers can represent a given song, so I want to reduce that super-redundancy to fractal coordinates.) ... The EZ-play-piano would not even require this, but just a statistical model of musical note sequences. ... Part of Melody Number 1 in my set-of-some-music, converted from an old tape recording, emitted by a "failed attempt to generate Noise" circuit... http://www.ultravires.net/notnoise.mp3 |
|
| |
//Every song that has been or ever will be composed can be played just by tuning the device// |
|
| |
In order for this to give you the *one* song you want, you would need to have an infinite number of monkeys tuning an infinite number of AMSPs. Then you would need a filtering method to figure out which monkey gets the prize. |
|
| |
The device you describe would in practice only produce new music. An AMSP creating an existing song would be a nearly infinitely improbable occurrence. This is magic, or a one-time miracle. |
|
| |
I believe that the AMSP would have four knobs and therefore an existing song will be harder to find than to find it on the radio. (The existing SMSP has only one knob.) I agree though that it will be impossible to get the lyrics right. But lyrics never make any sense anyway... I'd tune them out after the novelty of their meaningless random formants wore off. ... As for finding recognizable music, you know music when you hear it, that's why you don't tune a radio to a static frequency like a monkey would. (You wouldn't type Shakespeare like a monkey either.) |
|
| |
Lyrics could come from another more suitable process of generating them (speech synthesizer), and added to the musical waveform produced by the AMSP. |
|
| |
So is this a MIDI sequencer with a waveform generator attached then? The four 'co-ordinates' would then become just an algorithmic expression of the MIDI file being played. |
|
| |
Nope, it's a fractal waveform generator. My assumption of four knobs is based on an analogy between the mandelbrot-julia set conjugate which has 4 dimensions, hence 4 coordinates, called a quaternion. To play a fractal, that is, get an audio waveform from it, there are many methods. One is to simply move along any axis, starting at the point chosen by the coordinates, at an audio-like rate, say 44100 per second, and feed the number of iterations of each point into a digital to analog converter. The number of coordinates is totally arbitrary... you could generate a fractal in any number of dimensions, increasing the likelyhood of finding music waveforms, and then later eliminate the dimensions which do not affect the music much, or as I would say, do not intersect music-space... (music-space is where the-set-of-all-music is, and can be represented by a mathematical formula)... Someone asked me, "can I record songs on an AMSP? ... No, it would be like adding colors to the rainbow, they are all there already. How can that be?... Well a radio antenna for example may have signals for hundreds of songs in it already, but you only hear the one you tune into. You can tune them all at the same time, but it sounds awful, see Eigenradio! Imagine the AMS as a wave function defining music, containing all music, from which specific music can be extracted. ... The AMS player simply moves through musicspace, picking out the wave form of a song in real time and feeding it into a speaker. --- The AMS function could possibly be estimated by playing every known song into a large neural net which would not be limited to recalling only known songs but would have learned the fundamental essence of music, the fractal shape of musicspace, within which all possible musical waves would be bound. (If there was only one continent land mass, then any new city would certainly be built There.) --- My Some-Music-Set Player (see "fractal being played"), otherwise known as the "noise generator that doesn't work" is a simple circuit, far simpler than anything MIDI. It is one that neither oscillates nor reaches a stable state, from which I expected noise.
I work with a soldering iron, not high level protocols. |
|
| |
So you're basically saying - imagine a great big noise. An ultimate noise, if you like, which consists of all possible noises heaped together in one great big noise. Because this big pile of noise has all possible noises contained within it, then it must contain all possible music (songs already written, but also those waiting to be wrote) as well. And you can navigate this big heap of noise through 4 handy-dandy dials, like an acoustic four-dimensional etch-a-sketch... |
|
| |
I suppose if an infinite number of monkeys typed for long enough they'd eventually come up with this idea as part of the third act of A Midsummer Night's Dream... |
|
| |
Betcha can't make Wagner with a fractal algorithm. |
|
| |
Assuming that you strip music down to its most basic element, a melody, and composed a piece that lasted 10 notes, over a limited range of 2 octaves, with only whole notes, half-notes, and quarter-notes, there would be 13 X 10^30 possible combinations. Those four knobs must be VERY big. |
|
| |
BTW: Where did you get that first MP3 link? It sounds like my old Atari 2600 when I spilled orange juice into it. |
|
| |
The "big noise consisting of all possible noises" is right on except that all the noise has been excluded from the set and only the music remains. The etch a sketch analogy is correct too. And I've already said that human beings tend to tune to radio-frequencies that contain good music, whereas a monkey is likely to tune to static. Both the mp3 and the wav I linked came from a tape recording of my single-knob prototype, which I may link a picture and a medley of. It might be easier to search a fractal for Wagner with a computer than to turn the knobs and find it, but fractals were discovered because of things much more complex than music, such as the discovery that the coastline of Britain has infinite perimeter and is a fractal. Consider also the similarity between fractal landscapes and topographical audio spectrum analyzer displays. An existing .wav (like Wagner) CAN be algorithmically compressed. I think I could make a calculator chip play Wagner if I wanted to. |
|
| |
Try using the function A=(INT(T/8192) AND (T AND 8191))=0: T represents time as somewhere around say 8192ths or 44100th's of a second. "A" represents a waveform which should be integrated (turn up the bass). Also try plotting the set (X AND Y)=0. First, you will hear notnoise.mp3, even though this was NOT how it was originally generated, but is an algorithmic compression. Second, you will see the immediately recognizable fractal. Note: AND is the boolean function applied to binary numbers. When it evaluates to true (equals zero), the coordinates are in the fractal. --- The SMSP does not use the AND function at all, and only generates notnoise.mp3 when it's initial state is clear. ... also I had not emphasized that the SMSP generates interesting Visual and sound patterns, because I do not know whether the AMSP will be as interesting visually or not. --- THIS experiment was done in old fashioned BASIC because it is cheap, fast, and interactive. (you'll need to know either how to generate a sound file, or what the I/O port address of your digital to analog converter is). |
|
| |
I think it would be possible to generate an existing musical arrangement, i.e. the sheet music of a Bach concerto, however, I think it would be impossible to generate existing sound, i.e. Glen Gould playing a Bach concerto on a certain recording. It's my gut instinct that finding the undiscovered fractal called The-Set-of-All-Music is not realistic. In fact, it is my gut instinct that finding the undiscovered fractal called The-Set-of-Anything-Resembling-Any-Recorded-Music is not realistic, even if you hijacked SETI@home for this purpose for 50 years in all the parallel universes ever imagined by The-Set-of-All-Known-Star-Trek-Fans. With our luck, if they ever did find something it would be Jennifer Lopez. |
|
| |
What [xrayTed] said.
But wait a minute if we randomly burned 1s and 0s onto a CD maybe, just maybe we would end up with the new Justin Timberlake album ... |
|
| |
- well, maybe that's just a risk
you've got to accept. |
|
| |
I'm staying away from this (oops!) because I can't talk the science. Sounds like rubbish, but interesting to chew on for a minute. |
|
| |
There's no science here, snarf, it's all rubbish. Synopsis: there's this big noise which contains all possible noises within in, and you have these four magic knobs that you turn and 'extract' out all the songs that lie within. Everything else is fluff to make it sound impressive. |
|
| |
Music is supposed to sound impressive. Besides this is the ultimate in free music "sharing". Soon you can write something like f(z)=z2+c on your T-shirt with a marker pen and everyone gets all the legal free music they want. The proof is in the pudding. - Good inventions are not described in science books. Did the wright brothers study aeronautic engineering and get a pilot's license? |
|
| |
// the-set-of-all-known-star-trek-fans... // |
|
| |
This sounds similar to the tune that was played in the millenium dome. It was generated by a computer and intended to play for a thousand years without repetition. It seems possible that someone listening in a few years (it is playing in a lighthouse somewhere also) may hear 'Monkey Wrench' or 'Eine kleine nachtmusik" but about as likely as throwing dice all one's life without getting a six. |
|
| |
Also, what's the deal with
//the coastline of Britain has infinite perimeter//? Doesn't seem quite right somehow. |
|
| |
The AMSP uses the number 0.1234567891011121314151617...
which has all the other numbers in it. See monkey link. File may play as sound or read as text, but turn up the volume. Sound includes instructions for monkeys. |
|
| |