Computer recognition of V/CV speech units based on linear and non-linear dynamical system models using brain like computing and statistical learning algorithms