NOISE AND REVERBERATION REDUCTION FOR HEARING AIDS

Not Available Latent Semantic Analysis (LSA) is an automatic mathematical/statistical procedure for extracting and representing the semantic similarity of words and passages in any domain by analysis of large bodies of representative text. LSA has proven extremely effective at indexing and categorizing document based upon content. This Phase I SBIR research project will establish the feasibility of using LSA for the purposes of automatically filtering electronic mail messages. We will also examine the issue involved in customizing our algorithmic approach to the requirements of a cross section of military office environments. Based upon our results, we will design an LSA-based software system for automatically categorizing electronic mail messages into a set of user-defined folders, in a general purpose fashion that will support the use of this tool across a brad spectrum of potential customers.