Speech synthesis for glottal activity region processing