Abstract

Filtering is a basic operation, almost required in every signal processing system. The classical filtering is time-invariant, the sampling frequency and the filter order remains unique. Therefore it can render a useless increase of the processing activity, especially in the case of sporadic signals. In this context, adaptive rate filtering techniques, based on a level crossing sampling are devised. They adapt the sampling frequency and the filter order by analyzing the input signal local variations. They correlate the processing activity to the signal variations. The computational complexities and output qualities of the proposed techniques are compared to the classical one for a speech signal. Results show a drastic computational gain, of the proposed techniques compared to the classical ones, along with a comparable quality. It promises a significant processing power reduction of the proposed solutions compared to the classical ones. (C) 2013 Elsevier B.V. All rights reserved.