Databending - Creating and Listening to RTF Files

Last week Boing Boing had an interesting post about using EXE files to create found audio clips. The basic idea is to use an audio editing program such as Audacity to open and listen to raw data files as audio files. The process is called databending and has been around a good many years (I know, I'm always late to the party). Anyway, I played around with this a bit, randomly selecting program files from our computer at home to see what kind of audio they would generate. Uncompressed files seem to work the best since there's more data to "read", although shorter files can be copied & strung together to create repeating noise patterns. Compressed files (such as .jpgs) seem to yield mostly static.

I thought about this over the weekend and wondered if you could have some measure of control, based on the makeup of the original data file. I played around with this a bit more using MS Works word processor to create & export various file types (DOC, DOCX, RTF, TXT, HTML) and open them as audio files. I found that RTF (Rich Text Format) files work great because they contain a lot of data, especially if you use the paint function to draw something into your file (files that only contain text are still pretty small). Also, adding clip art can make for some interesting audio. The only problem with the audio from RTF files is a constant high frequency squeal, which can be reduced/eliminated using Audacity's noise removal feature.

So.... here's an example:

A four-page document containing clip art pictures of accordions and keyboards, saved and exported as an RTF file.

The resulting audio file with removal of high frequency squeal, otherwise unaltered and exported as an mp3 file:

Oddly, the same clip art document minus the two brown keyboards pictures yielded the same audio pattern, but at a higher pitch.