Our work covers the use of the artifical neural networks for speech separation using a cocktail party effect simulation. It uses a modifed Wiener filter that provides audio samples for subsequent modification using a Least-Mean-Squares filter. https://drive.google.com/file/d/0Bycfi0ce0B1mSG11bXlPa09EUlU/view