9 Royalty-Free Audio Tracks for "Neural Network"

Audio generated from training a neural network on violin sounds.

Results 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, sounds pretty lo-fi. The reason could be input compression needed for 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 wasn’t saved anywhere after they produce a thing. Global parameters (output compression, neuron count, drift rate etc. ) aren’t the same from sample to sample, too.

High-quality sound effect for relevant scenes in movies/videos/music, etc. Высококачественный звуковой эффект для соответствующих сцен в кино/видео/музыке и т. Д.

The results of an attempt at using a generative adversarial network to generate minimal techno, one measure at a time. The number in the sample's filename roughly corresponds to the training epoch where the sample was taken from.

A 120bpm amen break generated by a lstm recurrent neural network. Generated using theano-backed keras library in python, trained on the gpu. No post-processing applied.

Created via a neural network, science is great huh. Chiptune 8 bit song.

An example of an ugly generation to supplement the paper, transformer neural networks for automated rhythm generation.

An example of an bad generation to supplement the paper, transformer neural networks for automated rhythm generation.

Example of a good, from scratch generation to supplement the paper, transformer neural networks for automated rhythm generation.