User-interactive knowledge-based speech enhancement

Presents the details of a system which combines symbolic and numerical (signal processing) processing in a unified, interactive environment for the purpose of enhancing distorted speech and audio signals. The system is intended to allow the user to effectively and efficiently combine many different signal and symbolic processing techniques in order to interactively create an enhancement solution which is tailored to the specific distorted signal of interest. The primary goal is to allow the use of many different classes of available information (signal, noise, derived constituent structure, phonemic content, talker characteristics, segment boundaries, etc.) in the enhancement process, and to make the combination of the enhancement procedures easy for an experienced and expert user