作者: Vassilis Pitsikalis , Petros Maragos , Iasonas Kokkinos
DOI:
关键词:
摘要: In this paper, we explore modern methods and algorithms from fractal/chaotic systems theory for modeling speech signals in a multidimensional phase space extracting characteristic invariant measures like generalized fractal dimensions Lyapunov exponents. Such can capture valuable information the characterisation of - which is closer to true dynamics since they are sensitive frequency with attractor visits different regions rate exponential divergence nearby orbits, respectively. Further examine classification capability related nonlinear features over broad phoneme classes. The results these preliminary experiments indicate that carried by novel feature sets important useful.