Data Bending

What is data bending? Down at the intersection of electronic and experimental is an idea called data bending. It is the abuse of software to produce new sound. Software including programs, files and raw data of all kinds. Why do it? Good question. It's probably some mix of the following: (a) the joy of new and unexpected sound, (b) the possibility of taking data that has a particular meaning and translating it into sound - an image file, for example, (c) the overriding of cold, byte-perfect computer logic, (d) it's fun to fuck with things. There's one group of data benders I know of here.

A cousin of data bending is circuit bending - the hardware equivalent. Electronic devices of all kinds are hacked to produce new results. Again, there's a group here, and good links here, here, here and here, as well as many other sites.

One data bending idea is to damage CDs so they become data-impaired. Here is the story of my first attempts ...

My First CD Bends

First of all, what's the goal here? Is it to produce random skipping? Or to get other audio artefacts not sounding like the original? Damned if I know. I'd think the nicest thing to do would be to burn some of your own audio to audio CD for bending. I went looking for info on exactly how data is stored on a CD. Interesting articles here and here. None of the audio samples below have been edited except for a little volume adjustment.
 

1st weapon: Pizza cutter wheel
CD: Spoken word

Started with light scratching, more circumferential than radial. No effect. More scratching. And more. Wow. So much scratching, so little effect. Oops, the edge snapped into a split. Still plays fine. More scratching. Finally, some tiny artefacts; little digital blips, but still playing fine. Just a little more scratching, and now the CD won't play at all. Next please ...
 

2nd weapon: Box cutter knife
CD: Poetry/spoken word
 

Audio sample: BOX CUTTER 1 199kb, MP3, 22050Hz, 64kbps, mono   Audio sample: BOX CUTTER 2 229kb, MP3, 22050Hz, 64kbps, mono

I learned from above sites that CD playback is less sensitive to radial marks than circumferential marks. So, gradually built up radial scratches. Well, CD technology is looking quite impressive at this point; with easily 50 deep radial gouges, each covering the entire radius, the bloody thing still plays fine. After about 100 gouges, I got impatient and put on about 8 slightly curved scratches, and now it doesn't play at all. Hmm, I didn't think I'd need so much patience. I suppose CD players must be wired to be clever, so far as not continuing with a damaged disc. Next CD: using slightly lighter cuts this time. Now I get to a point where fast-forward is well screwed up; it stutters back and forth randomly, but the CD still plays fine. I must be close to something. Once again, there simply comes a point where the CD won't play at all. Boy, there's a fine if not non-existant line between playing and failure. Even if I could hit that point, it would be too fussy to reproduce on demand. For the record, I am trying all tracks at various positions. Next CD: Now I try smaller cuts that don't cover the entire radius, thinking this will help avoid total failure. Trying to use the fast-forward's difficulty to find the damaged areas. With dozens of small cuts, I get digital noise bursts that track to the audio, but not very impressive. More cuts and I finally start to get some tasty skipping; this is better! There's at least 100 cuts, about 1î long, on one quarter of the disc only, radially, in the middle. Next I add a couple of dozen more cuts, still 1î, over the existing cuts, at about 30 degree angles. Even better! Wild stuttering and skipping anywhere within plus or minus 2 to 3 tracks! It's so good I'm leaving this CD as it is in case I never repeat it.

 

3rd weapon: Steel wool
CD: Solo jazz drumming
 

Audio sample : STEEL WOOL 201kb, MP3, 22050Hz, 64kbps, mono

Started with light circular abrasions, again in a 1î wide patch in the middle of one quarter of the CD. No affect. Heavier scouring ... now all tracks from 6 to 14 (the end) won't play. But tracks 4 and 5 do some lovely random skipping of about plus or minus 15 seconds or so. If you listen to the audio sample, know that the original was a coherent and quite straightforward jazz drum solo. I'm starting to think the ideal tool would be a fine pointed abrasive ìpenî; there must be such things.

 

4th weapon: Sandpaper
CD: Rock music
 

Audio sample : SANDPAPER 841kb, MP3, 22050Hz, 64kbps, mono

Here I attempted to make thinner scours, in an arc covering one quarter of the CD, from near the outside edge to near the middle. One light swipe and I get great skipping, often within plus or minus a few seconds, sometimes between whole tracks. Best so far! Different start points yield different effects, including a very rhythmic pulsing. I pasted some parts together for the audio sample.

 

5th weapon: Marker pen (Paper Mate permanent W10 red)
CD: Spoken word & accordion (French lessons!)
 

Audio sample : MARKER 670kb, MP3, 22050Hz, 64kbps, mono

Following on the success of the arced scour, I started with one arced line, written over a few times and dried. Wow, it sends it haywire. More skipping than ever. Probably the best of all, since markers can be applied with greater control and subtlety.

 

6th weapon: Luck, nature, the Gods
CD: Lens cleaner (Memorex CDL-100)

Audio sample : LUCK
677kb, MP3, 22050Hz, 64kbps, mono

This CD appears spotless, yet skips very smoothly and creatively; it always has. Perhaps the lens cleaning brushes are the reason. It's a natural born bender. After the spoken part, you get "Carmen Remixed". The best sounding results, but alas, not reporducable on other CDs. Yet.
 

Conclusions

Clean cuts, regardless of depth, are very difficult to balance between effect and total failure. Any type of damage is best kept well inside the data suface; away from the edges. And playability seems more solid when less than 360 degrees are marked. Scouring is better than cuts. Thin, arced bands of marking work better. Marker pens show great promise. Further to my first comment, what more could one expect to achieve doing this? As suggested in the Yahoo group, I will be trying the microwave for the sheer fun of it. And perhaps refining the best of the above techniques.
 

UPDATE

I completed a piece for a remix competition, which you can hear here. I was hoping to be able to match music tempo to skip tempo, but that seems more difficult than worthwhile. Skipping tempo does vary from the centre (earlier) to the outer edge (later) in the range of about 60bpm to 200bpm respectively. But it's too hard to isolate skipping at the relatively short span that will match the music tempo. As for applying damage, the permanent marker seems the best all round. Very easy to apply marks with subtlety, and easy to re-arrange; q-tips with a dab of alcohol are perfect.