Corpus as model for a modular approach to (electro)acoustic composition...this is the main theme for my current work, both research and composition. It is based on both the theories and implementations of CataRT, and a more general application of the same concepts.
Adaptation of CataRT to Jamoma Modules...In the first few months of 2008, I took an interesting little piece of software (actually a large set of Max patches) and reworked them to form a series of Jamoma modules. The purpose was two-fold: I wanted to understand all the parts of a complex system, I wanted to actually use the system, and I wanted to modularize the whole thing in order to build on it. In the first phase I have essentially rebuilt the original set of patches into modules and made it so that I could control everything interactively to record tape parts or to make live electronics. In phase two, I will finalize the two solo string pieces using modular patches. Phase 3, running concurrently with phase 2, will see a fully modular set of corpus-based sampling components that will become an integral part of my general corpus-based project (see above).
Tracking voices speaking texts...in 2007, I used some of the ideas I developed for the "Maps" project, but applied them to tracking/processing speech samples. What I wound up using in the tape part for "2 Dreams" was voices (reading lists of words). I used some of the FTM library objects in Max to cut out grains from the voices so that various regions of a buffer would be left out (silenced) based on thresholding some parameter (where every frame generates a parameter that is judged with respect to some threshold and the grains whose values pass the test are output). In addition to a "voiced-unvoiced" criterion, I experimented with a "noisines" or a spectral flatness measurement. The result is speech that can loose its meaning but retain 'cadence' or some speech-like sound qualities, depending on how one defines the threshold conditions for the parameters. The majority of the tape part for 2 Dreams uses cut up recordings of people speaking.
Tracking string timbres...Since 2006, I have been working on a group of compositions ("Maps") for solo string instruments and Max/MSP that employ various methods of timbre tracking (in addition to envelope-tracking and onset-detection). The work has progressed though a number of stages. At first I used Miller Puckette's fiddle~ object. Later I used a derivative (Tristan Lejan's analyser~ object/suite of objects). As I experimented more, I disovered IRCAM's FTM Library, which brings with it a whole host of useful classes and methods for dealing with spectral information...and in a granular way that really appeals to me (see below). I used timbre-tracking in my two recent solo pieces entitled Maps.