Many of the other explorations described in these pages use mathematical systems of one sort or another to generate a datastream puportedly imbued with qualities observed in natural phenomena; which datastream is then mapped into a musical/graphical space. Scraper is an experiment from a different direction: the datastream is taken from something that already exists and is readily obtainable: pages from websites.
Websites consist primarily of digital documents written in HTML, and usually embedded with snippets of other languages, both artificial and natural. Insofar as HTML documents and the content contained therein can be understood from an information-theoretical point of view as being information-rich, structured and hierarchical on multiple levels, they would seem to be interesting candidates for input into an algorithm which churns out sound by "scraping" through raw data, interpreting it as musical code.
Scraper currently runs on Mac OS-10.7 or higher. You may enter any URL, or a search term for Google. The returned HTML page is displayed as raw text, and then begins to play, using the text as a sequence of instructions into a music synthesizer. Switching between levels looks at the HTML text from different hierarchical levels.
The two videos below were created as a demonstation. The HTML text is the home page of the online edition of the New York Times from Sunday, March 3, 2013.
Signals and Noises