It is shown how two-mass models for the vocal cords can be implemented efficiently with digital signal-processing techniques. This facilitates their use in speech-synthesis hardware. An example of such an implementation and the glottal pulses obtained with it are presented.