LSP application in speech enhancement and isolated word speech recognition
MetadataShow full item record
In this thesis, line spectral pair (LSP) representation is used for speech enhancement based on Wiener filtering with spectral constraints and is compared with the Forward/Backward adaptive filtering approach. The LSP representation is also used in isolated word speech recognition with a conventional dynamic time warping (DTW) approach. Then, the isolated word speech recognition using linear predictive coding (LPC) representation and LSP representation with Fuzzy Logic is introduced.