Compressed 96kbps MP3.
Spectrogram of complete composition (0-11khz grayscale); time-domain
waveform overlaid (blue):
Compressed 96kbps MP3.
As part of my investigations into the relationship between improvisation and phrase transformation in composition, this piece is based on a real-time processing patch built using Max/MSP:
A set of simple improvised motifs are played into by the composer, but are not heard. Instead they are subjected to phase vocoding analysis, and the results recorded into a sample buffer at a linear rate. The contents of this buffer are continuously played back at non-linear rates and converted back into a amplitude/time waveform in real-time, thus pitch and spectral content of the motifs are maintained, while temporal content of the motifs is distorted. The resulting audio signal is played back to the composer, and mixed with the composer's motifs as the input signal, with a band-pass filter attenuating this feedback. After much experimentation I found that a sinusoidal playback rate provided wonderfully rich musical development of the input material.
The net effect of this process is a ‘folding over' of the performed motifs. Because of the sinusoidal playback rate, certain parts of the motif are elongated, whilst others are shortened, and the motifs might be split into component micro-gestures and repetitions. Because of the feedback, certain parts of the processed motifs may repeat more than others, disappear for some time to return later considerably re-arranged.
With practice, I learned how to influence the sonic results by controlling the playback rate, buffer segment, envelope and feedback parameters. I found the mixture of generalised predictability and particular indeterminism a very fruitful form to generate musical phrases, and developed the macro-structure and progression of the piece accordingly.