Realization of a vowel-plosive-vowel transition by a tube model

Tube models, realized by lattice filters, can be used for the production of speech signals, since the model simulates the propagation of plane sound waves through the vocal tract. In this contribution the model contains a time dependent glottis coefficient. The parameters of this time dependent model can be estimated by minimization of an error, which describes the spectral distance between the tube model and the speech signal. The estimated vocal tract areas are comparable to areas, which are obtained by NMR or X-ray investigations. A procedure is proposed to model the voiced transition from a vowel to a plosive and to another vowel (VPV). This is realized by a vocal tract area transition of the tube model. The areas of the constriction of the plosive and the areas of the voiced sounds are estimated from speech signals.