Freesound - neural network → noise by arseniiv

NB: Spectrograms of several sounds are curious.

UPD: Here’s a description of the algorithm. (Let me know if it’s unclear.)

These result from running randomly-generated neural networks (all weights in neuron connections start random and then slowly drift when generating). Although sample rates used are 44.1 and 48 kHz, it sounds pretty lo-fi. The reason could be the input compression needed for a network to actually work. Each sound channel is an output from two separate neurons in the network. Each sample in this pack is generated by a separate network, as they weren’t saved anywhere after they produced a thing. Global parameters (output compression, neuron count, drift rate etc.) aren’t the same from sample to sample, too.

digital