Model based Binaural Enhancement of Voiced and Unvoiced Speech

摘要

This paper deals with the enhancement of speech in presenceof non-stationary babble noise. A binaural speech enhancement framework is proposed which takes into account boththe voiced and unvoiced speech production model. The usageof this model in enhancement requires the Short term predictor (STP) parameters and the pitch information to be estimated. This paper uses a codebook based approach for estimating the STP parameters and a parametric binaural methodis proposed for estimating the pitch parameters. Improvements in objective score are shown when using the voicedunvoiced speech model in comparison to the conventional unvoiced speech model