Design and implementation of an ideal Hilbert transformer using neural networks

Abstract

A method for approximately realizing ideal Hilbert transformers is presented. Ideal Hilbert transformers with two-sided impulse response can be obtained by the parallel connection of causal and noncausal digital filters using time reversal techniques. A noncausal filter is simply designed from a causal filter. The authors have designed and implemented an ideal discrete Hilbert transformer by the use of the proposed method, and give a few applications to digital signal processing. >