We present a revised flow model for voiced sound production which take
s into account the movement of the point at which how separation occur
s in the diverging part of the glottis. Although based on the same ass
umptions, this theory presents some important differences with the com
monly used Ishizaka and Flanagan flow model. A validation of the propo
sed model, based upon comparison with unsteady-flow measurements on in
-vitro glottal models, is presented. The measured pressures and veloci
ties as well as the flow visualizations are discussed in terms of thei
r consequences for glottal flow modelling.