Self-Attention - violin, machine learning audio
2021 | 7 mins
Performed by Darragh Morgan
Commissioned by Music Current/Dublin Sound Lab
Music Current Festival, Smock Alley Theatre, Dublin
9 November 2021
⁂
This work attempts to re-embody the output of a machine learning system that has been trained on music for violin. The performer becomes a nexus for seemingly contrasting technologies: notation, the body, and machine learning neural synthesis. A potential skeuomorph of the biological and the digital, Self-Attention instead treats technology as the external organs of the body, blurring the lines between subject and object.
Combing through the output of a neural network trained on violin music, a couple of musical fragments and gestures jumped out at me. And the thought can one call something the neural net generates a gesture, if it has no concept of an embodied technique and practice? Maybe such gestures originate in the music it was trained on, but maybe not. And does it even matter? So Self-Attention seeks to re-embody this music. Tuning disparities in the piece may underline the distinction between the embodied and the disembodied. Maybe they’re just character quirks.
