Piecewise self-affine fractal modelling of linear prediction excitations in speech compression

A technique for modelling speech excitations using fractal interpolation is presented. The fractal interpolation techniques used are based on iterated function systems (IFS). Two iterated function system models are used: the self-affine model and the piecewise self-affine model. The self-affine model is found to be inefficient for the purpose of representing excitations. The piecewise self-affine model is found to give a better model for the speech excitations. A low bit-rate speech coder is implemented using the piecewise self-affine fractal model.