Toy U-NET

This is a implementation of a basic U-NET Convolutional Neural Networks over a dummy dataset of sines mixed with white noise. The porpose of the sistem is to denoise the audio signals in the STFT amplitude domain. The objective of this repository is to serve as a basic usage example of the Pytorch framework to develop NN for audio processing.