Enhancement of cleft palate speech using temporal and spectral processing