The following article is by Jarmo Niinisalo.
A long time ago—in the summer of 2001 if I recall correctly—I ran into an intriguing page (Chaosmachine’s more recent page can be found here). The original page, which sadly resides only in the web archives nowadays, was all about this hidden “demon face” in one of Aphex Twin’s tracks, "#2" (the long formula) on Windowlicker.
This face was supposed to be viewable with a spectrograph program, so I decided to try it myself. A spectrograph basically visualizes the sound spectrum.
First I needed to extract the track from the Windowlicker CD, which was easy with CDex. The extraction of the whole track was not really necessary because the “face” is situated at the very end of the track, starting from the 5:27 mark and lasting for about 10 seconds. There are other “audio images” on this particular track as well (and one at the end of the first track), but the face is certainly the most exciting of them all.
After I had the wav-file, I used a program called Spectrogram to visualize the file. To my amazement, it worked, and I was soon staring at the “demon” face:
The alleged demon face
I was, however, not content with this. Why would the good ol’ Twin have added a “demon” face into one of his tracks? I mean, the man does have a weird sense of humor, but I always thought demons were more up BoC‘s alley…
While examining the image, I came to the conclusion that something was not right. So I started messing around with the settings of the spectrograph program, and after a bit of knob twiddling the mystery revealed itself: the face was supposed to be watched with a logarithmic frequency scale, not with a linear scale.
A linear scale provided the “demon face”, but with a logarithmic scale the end results were quite different:
The plot thickens
Why, it was none other than the Twin himself all the time, complete with his patented grin!
The settings which I used to get the above image were roughly the same as in the picture below (they aren’t the exact same settings since I’ve lost them somewhere, but if you try them yourself you’ll find that they are more than close enough):
Spectrogram settings for reproducing the face
After this “amazing” discovery, I contacted the guy at chaos.yerbox.org and informed him of the results. He seemed interested to add this new-found info to his pages, but I suppose he never got around to actually do it.
Next, I decided to inform the good folks on the IDM mailinglist about The Face. It turned out that this “picture to audio” -thing was really not hard to do at all and there was a Windows program called Coagula that could transform any picture into soundwaves with minimum effort. Aphex Twin himself had used a Mac program called Metasynth to do his images.
I tested Coagula and found it easy enough to use. Now this new discovery raised a new question: Were there more examples of this audio imaging available?
Luckily the soundscapes that Coagula spits out are quite easy to notice in their aural form, so I didn’t have to look for long before I found another example: On Plaid‘s “Rest Proof Clockwork”, the track “3recurring” contains a continuous stream of “threes” (as seen on the cover art of their previous album, Not For Threes).
Besides the Plaid discovery I couldn’t find any more examples of audio imaging on any of the CDs I own, but I didn’t bother to look that hard either. However it is clear that this audio/image stuff was “hip” at 1999, since both of the aforementioned records were released around that time.
As a nice finish, here are is a small gallery of some audio images from Aphex Twin and other artists:
The “hand” found on “My Violent Heart” from Nine Inch Nails album Year Zero