Técnicas seletivas de filtragem adaptativa de sinais de áudio degradados por ruído impulsivo.

Abstract

This dissertation presents, implements and compares some selective adaptive techniques of impulsive noise reduction on degraded audio signals. The systems are composed of two steps: detection and filtering. The detection step is based on the calculation of the derivative of the signal envelope and comparison with a threshold predetermined, determined from the characteristics of the signal. The filtering step, in turn, is based on progressive, regressive linear prediction and double. A post-processing, based on the Lagrange polynomial interpolation, increases system performance, providing greater gain in the signal-to-noise ratio (SNR). The results obtained by the gains in the SNR of the processed signal and confirmed through informal listening tests were shown to be quite satisfactory, since they present a considerable improvement in signal quality.