A New Hilbert Transformer Based on Parametric Wavelet Kernel: Application to Analytic Signal Decomposition of Speech Sub-bands

Abstract

This paper aims at providing a fast Hilbert transformer based on a parametric wavelet kernel. The used kernel is of the same form as that employed in speech decomposition into critical sub-bands to which the Hilbert transformer is then applied to carry out analytic signal decomposition. This is accomplished by utilizing a shifted half-band filter which has a parametric formula and can easily be adapted to any desired effective bandwidth.