Manifesto in Performance Captioning

I now have a clearer idea of what I want to accomplish this semester. I began studying memory and narrative so that I could bring these ideas to the work I have been a part of in sosolimited. I was also concerned with increasing my skills with mobile hardware. Merging these two goals has been problematic. Here are my proposals for two specific projects. One explores narrative, the other memory, but they both come together in a live performance.

Narrative project: Text Sampling

Earlier in the semester I was designing a reading machine that would try to remember a key phrase in a nearly endless stream of words. This machine used an algorithm to rate the phrases it encountered and saved what it believed was the most interesting text. My original sketch of the piece was limited by the predictablity of this algorithm and its inability to find actually meaningful phrases. Furthermore, there was no real connection between the work and the audience, and there was no compelling reason to watch the machine in action.

Addressing these shortcomings, I am proposing a change in direction. Rather than try ot have the microchip pick the relevant phrases, the user should be able to do so in real time. I am interested in converting this reading machine into a text-based sampler/sequencer - kind of a silent MPC-2000. The input and output are text, but like the MPC, the user is composing with existing streams of data. This sampler will specifically be used in live performance.

I will begin by using the Closed Captioning from a television signal as the input stream. By changing the channel, the user can control what is entering the system.

Sampling: The ability to reecord a phrase on the fly
A unique story can be told by selecting specific parts of a text. In fact, given enough input text and time, any story can be told. The relation between the new story and the original text becomes interesting and adds to the narrative. Once sampled, these clips can form part of a library.

Sequencing: Organizing these pieces in terms of how they connect to one another and how they loop.
How will the samples form a composition? Will it become a static graphic layout or something that is changing with time? In what ways can a text perform?

Because this project deals with text, it needs to address that fact that there are many additional layers to the medium. It should have some functionality that we don't find in an audio sequencer/sampler.

1. Setting key words and searching for them
2. Counting occurances of words
3. Associating a word with its neighbors
4. Archiving the incoming text so that it can found later

These additional features will help the user establish themes within the narrative. You can link remote parts of a text or to connect the incoming stream of text to previous parts of the performance.

Memory project: Visual Recording

I am interested in a leaving physical traces as part of a performance or as the result of a performance. A recording can be made of a live audio performance and this recording can be played back at some time in the future. I am concerned with finding a way to record a similar performance in text so that the results of the performance can be saved in some way.

As I develop the text sampler, I want to keep in mind this ultimate output. This device will most likely be some kind of printed, but could be abstracted in a number of ways. It doesn't have to print the words themselves but could look at characteristics of the narrative.

Giving a physical quality to both the performance and the output has become a goal of the semester. I feel that I'll be in a better position to think about this output stage once I have successfully created the text sampler and performed with it. I plan to keep this ultimate goal in mind and return to it when I have made significant progress with the first project.