In Phantom Limb, I employed a compound sound synthesis approach. Four types of sound synthesis models are used: an extended form of dynamic stochastic synthesis, subtractive synthesis, additive synthesis, and physical modeling. The algorithms for sound synthesis and and generative musical structures are implemented in the Supercollider programming environment.
The synthesis algorithms are also controlled by the simulated body structures using mass spring systems and artificial neural networks.This neural network can exhibit
recurrent connections and signals propagate with time delays. The activity of the neural network can affect the properties of the mass-spring system and vice versa.