It is known that speech signals posses some fractal properties. A novel approach for speech modeling is therefore by means of fractal models. However, when speech is neither self-affine nor self-similar, suitable fractal models, such as the piecewise self-affine ones, should be used. In this paper we describe an optimal determination of the parameters of the piecewise model using Genetic Algorithms. A proper tuning of the algorithm was first performed. Then, different types of constraints on the search space were analyzed and the best trade-off between SNR, compression ratio and computational complexity was found. Finally, the fractal models estimated with GA was applied to speech signals. Some results, namely original and synthetic signals and segmental SNR, will be reported.
Genetic optimization of piecewise self-affine fractal interpolation with application to speech modeling
MUMOLO, ENZO
1994-01-01
Abstract
It is known that speech signals posses some fractal properties. A novel approach for speech modeling is therefore by means of fractal models. However, when speech is neither self-affine nor self-similar, suitable fractal models, such as the piecewise self-affine ones, should be used. In this paper we describe an optimal determination of the parameters of the piecewise model using Genetic Algorithms. A proper tuning of the algorithm was first performed. Then, different types of constraints on the search space were analyzed and the best trade-off between SNR, compression ratio and computational complexity was found. Finally, the fractal models estimated with GA was applied to speech signals. Some results, namely original and synthetic signals and segmental SNR, will be reported.Pubblicazioni consigliate
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.