Audatic is a system to intelligently modify sound using state of the art deep learning technology and unique datasets. The personalized sound environment allows users to customize sounds to their individual taste. Applications include effortless interactions in noisy places (especially for people with hearing impairments), and realistic audio-environments for augmented reality. In order to create real-time systems that run on mobile devices, we are combining cutting-edge algorithms for efficient implementation of deep neural networks with novel deep learning techniques for detecting, filtering, and modifying sound.