Implementations of algorithms to minimize the border effect in discrete wavelet transform

URIZ, A. J.; AGUERO, P.; HIDALGO, R.; CASTIÑEIRA MOREIRA, J.

Bariloche

Congreso; XV Reunión de Trabajo en Procesamiento de la Información y Control; 2013

Universidad de Rio Negro - Sede Andina

In speech processing, one of the first steps for the analysis of the   signal is to reduce the noise that it contains. One of the most used   noise reduction methods is the Discrete Wavelet Transform (DWT). The   main advantage of this method is that it can operate with non-stationary   signals. On the other hand, its main disadvantage is the amount of   required data memory which difficults its implementation in a low cost   digital signal processor (DSP). Also, discontinuities may appear in the   border of each processed frame due to the finite number of samples used   to compute the DWT. The main goal of this work is to compare a set of   implementations of noise reduction algorithms based on the DWT, which   also are implemented in a dsPIC of Microchip, in order to assess their   capabilities to minimize the before mentioned disadvantages. Objective   ans subjective experiments are carried out to compare the performance of   the methods, and experimental measurements are done to analyze the   performance of the algorithms implemented in the dsPIC.