Remezアルゴリズムに基づくShort Time DFTのデシメーションフィルタの最適化

Abstract

Decimation filters of the ST gDFT Hilbert transformer are examined through frequency response with employing Remez algorithm. Optimization is performed alternately on the frequency and time domain with restricting filter coefficients. This optimized response of the Hilbert transformer also defines the optimized weighting function for the decimation filters. In other words, the optimized Hilbert transformer response based on the Remez algorithm deduces the optimizing weighting function of the significant decimation filter in the ST DFT.