Models (in the style of ?) type beat is a prototype of a performance that deals with mimicry and simulation, conjuring the auras and moods of pop through a series of artificially synthesized vocal invocations. It reworks and recreates multiple “deepfakes” through the cloning and simulation of synthetic human vocalizations generated by deep neural networks trained with several artists, to create a kind of “metaperformance.”

In essence, these entities do not come from any era, they are not a specific person, they have not lived, and they will not die. Models is a continuation of the themes and questions explored in Lee Gamble’s audio essay Face2Face (2020)—a narration of an “imperfect” stage in the training process of a neural network, a vision of a socioculturally cloned world, and more an absurdist pantomime than a demonstration of the unsettling mimicry of human artistry that these networks will soon imperceptibly replicate.

Thanks to:
Joseph T. Colonel – Zoom discussions, Vocal mode! training, machine learning, artificial intelligence design and programming; Ben Hayes, Charalampos Saitis, Gyêirgy Fazekas – (NEWT) Deep Neural Network-Based Singing Voice Synthesizer.