Parametric Analysis of Speech Signals Based on Estimation of Joint Source-Filter Model Using Evolutionary Computation

摘要

This paper presents an analysis-by-synthesis algorithm based on joint estimation of speech parameters. The main advantage of the proposed algorithm is that both vocal tract and glottal source parameters are estimated simultaneously in an automatic way. The use of evolutionary algorithms is proposed to optimize these parameters. The results show that this strategy seems to be feasible for some applications such as compression of speech signals and voice conversion.