A [see page 17, model] for generating speech using a pulse-generator and a noise-source which is passed through a vocal tract filter.