April 27, 2002 12:58 AM   Subscribe

Hrm. Aphex Twin -- visual artist (via memepool)
posted by delmoi (18 comments total)
Oh. My. God. Richard D. James is my fucking hero.
posted by sixfoot6 at 1:45 AM on April 27, 2002

Could be that Richard D. James is running a programme to convert images to sound using something like MetaSynth - what you are seeing here is the reverse process, uncovering the images he has used.

But on the other hand, ATs music is so distorted, fucked and filtered that images are likely to emerge from the multitude of levels...
posted by mook at 2:00 AM on April 27, 2002

There's a better image here, which makes it fairly clear that this was not accidental.

If you're interested in hearing the sound that produces that spectrograph, you can download a fairly large WAV of it here. Unfortunately you can't MP3 it - the MP3 compression removes the image.

Having said all that, I still hate his music!
posted by Mwongozi at 4:29 AM on April 27, 2002

I think it looks like a sinister E.T.
posted by dabitch at 5:11 AM on April 27, 2002

Yup. I heard about this a long time ago. It's purely intentional. Wish I could remember where I found out about this. It's being a real pain finding anything via Google, but he's been mentioned as Using Metasynth several times.
Coagula is also usable for this kind of processing.
posted by Su at 5:18 AM on April 27, 2002

This is very, very cool. I hadn't hear about this but really like his music. Nice one.
posted by nedrichards at 7:21 AM on April 27, 2002

Unfortunately you can't MP3 it - the MP3 compression removes the image.

So I guess this is some really clever scheme Richard D. James came up with so people would have to buy the cd to get the images. Clever.
posted by bobo123 at 9:44 AM on April 27, 2002

It's worth buying just so you can get the video thats on the cd. Classic stuff.
posted by atom128 at 9:50 AM on April 27, 2002

me like this much too
posted by azazello at 10:12 AM on April 27, 2002

happy happy happy
i wish my face looked liked that when i wanted it too...

That second image is twisted.
posted by Espoo2 at 11:18 AM on April 27, 2002

Come to daddy!

I've never heard of "reverse-engeneering" performed on music, but this must be the first. It just goes to show how ingenious Richard's D. James' music really is.

I'm going to listen to drukqs again.
posted by Down10 at 1:13 AM on April 28, 2002

i bet it is the likes of MetaSynth or something similar, at first i thought that this was what the entire track sounded like, but that its just a 9 second, while cool, inst mystical or mysterious.

i think im gonna listen to drukqs as well.

oh, buy a venetian snares album while youre at it.
posted by Satapher at 2:51 PM on April 28, 2002

It just goes to show how ingenious Richard's D. James' music really is.

Yeah, it's real ingenious to use an off-the-shelf software product that is specifically designed to generate sounds from pictures for, well, what it was intended for.
posted by kindall at 4:12 PM on April 28, 2002

This is very different -- and much more impressive -- than just "dinking around with off-the-shelf software". Metasynth uses images as an input, and generates sounds based on those images -- but the correlation between the sound and the image are completely arbitrary, and wouldn't make sense in any other software. (Example: red pixel = stereo pan right, green = pan left). A cute trick, but the sound generated doesn't really have anything intrinsic to do with the image; you could use a different set of rules and end up with a completely different sound from the same image. If you were handed the audio file and didn't know the arbitrary rules that metasynth uses to convert pixels to sound, you'd never be able to retrieve the image.

The images Richard James is generating are actually intrinsic to the sound itself, using different densities of various frequencies to create an image: if you look at a spectrogram of the sound waves using *any* software, you'll see the same image. There aren't any arbitrary rules to be followed to see the image; just map frequencies vs time and there it is.

Which is pretty fucking ingenious. Sort of like steganography, but much trickier, since this would have to be audible: he would've had to compose the music so that at least the rough approximation of frequencies he'd need would be present in the right places (big blast at all frequencies on each side of the head, little high-frequency blips for the eyes, a persistent band at a lower frequency where the mouth is going to be, etc), then go back and use fourier transforms to fill in the finer detail of the picture. I don't think you're going to manage that with off-the-shelf anything.
posted by ook at 7:49 PM on April 28, 2002

Just realized I used " marks around what was not a direct quote. Sorry about that.
posted by ook at 7:53 PM on April 28, 2002

word, ook.
posted by Satapher at 9:29 PM on April 28, 2002

MetaSynth is an additive synthesis program and in it, vertical axis = frequency, horizontal axis = time, and luminosity = amplitude. This method of operation means that whatever picture you paste in will show up in the frequency spectrum of the sound, since that is merely the reverse of the process used to create the sound in the first place. The other cool stuff (hue = stereo pan, etc.) will of course not be interpreted by a spectrum analyzer, but the basic shape will come through visibly enough.

He just pasted a picture into MetaSynth, rendered the sound, liked what he heard, and used it on his record. Big whoop.
posted by kindall at 10:44 PM on April 28, 2002

Now that I've downloaded and listened to the clip....

You're 100% correct. How disappointing.
posted by ook at 7:21 AM on April 29, 2002

« Older Terminal Error   |   'Cambridge Students Beaten by Israeli Army' Newer »

This thread has been archived and is closed to new comments