The goal of this project is to apply deep learning techniques to digital signal processing. Potential applications are "de-clipping" signals that were recorded too loud, "de-reverbing" sounds that were recorded in echo-y spaces, or removing unwanted sounds from recordings of the human voice.