So I settled for this trippy video processor that takes video and audio, feeds the video through some delay lines and then convolves them with randomly generated kernels weighted towards specific effects, based on thresholds in the audio signal. I figured it would be fine, since people at raves don’t usually tend to be too picky.

For the video, I used one of my favorite songs, Figure 8 by FKA Twigs, paired with one of my favorite dance clips, an excerpt from the pas de deux of Matthew Bourne’s all-male Swan Lake. I hope you enjoy!

In this well-known piece Alvin Lucier elegantly demonstrates that any acoustic space is a system that processes sound.  In a room, sounds bounce off walls, ceilings, floors, and other reflective surfaces, producing standing waves.  These waves add harmonics and reverberation to the original sound.  The frequency of the added harmonics depends upon the distance that the sound waves travel between the various reflective surfaces (amongst other factors).  In many rooms the added harmonics and reverberations are so subtle that they go unnoticed – we don’t think of sounds in a room as being processed.  But in this piece Lucier reveals these subtle transformations in a very dramatic way.

In I am Sitting in a Room, the composer reads a text aloud whilst using a microphone to record his voice.  The recording is then played back in the room, and the microphone now records the sound of the playback occurring in the room.  This second recording has now gone through the room twice (once during the original reading and again during the playback), and the harmonics and reverberations from the room are accumulating.  The second recording is then played back and re-recorded, and the process repeats over and over.  With each iteration of the process more and more reverberation and harmonics accumulate, until eventually the composer’s voice is an unintelligible blur of gorgeous room-o-licious tone-clouds.  Glorious!

This piece also demonstrates the use of feedback in a system, but on a time-scale that is larger than what we might expect.  Feedback is used in filters and echo systems to produce effects that are immediately noticeable.  In this piece, the feedback accumulation is rather slow (with one cycle of the system feeding back once every 90 seconds or so), allowing us to hear the subtle transformation of each feedback cycle with much more detail.

In fact, I would say this piece is a prime example of feedback art.  Aside from generating the original source signal (the composer reading a text), the only procedure that is used in the work is to present the output of a system back to the input stage of that same system.

Here is a live performance of the piece:

And here is an interesting homage called “I am Sitting in a Video Room”:

This piece also demonstrates the slow accumulation of feedback through a system.  In this case the system is not an acoustic space, but is Youtube.  A video is uploaded to Youtube, resulting in compression artifacts, little glitches in the sound and image, that are usually not noticeable.  The Youtube video is then downloaded and re-uploaded, resulting in a second iteration of artifacts.  The process repeats and the artifacts accumulate until the images and sounds in the video are a computery scrambled-up nightmare.  Terrifying!

And then there’s this:

Mind = Blown.

While not a direct reference to Lucier’s piece, Search by Image, Recursively, Transparent PNG, #1 by Sebastian Schmieg is another great example of feedback art.  The artist starts off by uploading a blank image to Google’s search-for-similar-images service.  The first hit that results is then presented back to Google and a similar-image search is performed on that image.  Each iteration of the system produces an image that is slightly different from the last one*, creating a slowly transforming visual landscape.

• This feedback system could have resulted in a recursive loop rather than the linear progression we see in the video.  If a similar-image search for image A produces image B, it makes sense that a similar-image search for image B would produce image A.  This would create a video that just goes A-B-A-B-A-B-A… To avoid this the artist skipped search results that had already been used in the video, choosing the next-most-similar image beyond the already-seen image.