Abstract
Aerodynamic simulations of /aCa/ utterances were made using a low-frequency model for upper vocal tract airflow and a two-mass model for the voice source. These simulations helped increase insight into the results of an empirical study of flow during running speech. The various sources of flow, including wall compliance, were examined for their contributions to total flow from the mouth. The two-mass model was modified to allow for more natural glottal flow during abduction and adduction. Even with modifications the two-mass model was not sufficient to model source variations during running speech.