Abstract-For voice rehabilitation after total laryngectomy a shunt valve is usually placed in the tracheo-esophageal (TE) wall, thereby enabling the production of a TE voice. Some patients, however, are unable to produce a voice of sufficient quality. Furthermore, the TE voice is low pitched, which presents a problem especially for female laryngectomized patients. The voice quality after laryngectomy might be improved by introducing a voice-producing element (VPE) into the TE shunt valve. In this study a sound generator was developed that is suitable for application in such a VPE. This sound generator consists of two elastic membranes placed parallel inside a circular housing. A substitute voice source is created when the membranes start to vibrate via a constant flow of air passing between them. To determine the optimal membrane configuration for proper functioning under physiological conditions, up-scaled physical VPE models with different membrane geometries were evaluated using in vitro experimental tests. For certain membrane geometries the tests showed that a basic sound, containing multiple harmonics, could be successfully produced under physiological air pressure and airflow conditions. The fundamental frequency (60-95 Hz) and sound pressure level (57-78 dB, at 15 cm microphone distance) were regulated via changes in the driving pressure, thereby enabling the possibility of intonation in laryngectomized patients' speech. The obtained frequency range is considered appropriate for producing a substitute voice source for female patients. The geometry considerations in this study can be used for the development of a true scale VPE that can be evaluated clinically, to eventually replace the voice after laryngectomy.

A computational model for exploring the design of a voice-producing voice prosthesis, or voiceproducing element (VPE), is presented. The VPE is intended for use by laryngectomized patients who cannot benefit from current speech rehabilitation techniques. Previous experiments have focused on the design of a double-membrane voice generator as a VPE. For optimization studies, a numerical model has been developed. The numerical model introduced incorporates the finite element (FE) method to solve for the flow-induced vibrations of the VPE system, including airflow coupled with a mass-loaded membrane. The FE model includes distinct but coupled fluid and solid domains. The flow solver is governed by the incompressible, laminar, unsteady NavierStokes equations. The solid solver allows for large deformation, large strain, and collision. It is first shown that the model satisfactorily represents previously published experimental results in terms of frequency and flow rate, enabling the model for use as a design tool. The model is then used to study the influence of geometric scaling, membrane thickness, membrane stiffness, and slightly convergent or divergent channel geometry on the model response. It is shown that physiological allowable changes in the latter three device parameters alone will not be sufficient to generate the desired reduction in fundamental frequency. However, their effects are quantified and it is shown that membrane stiffness and included angle should be considered in future designs.

Functional restoration of the voice after laryngectomy with a double-membrane VPE appears to be a feasible concept for female laryngectomized patients with a hypotonic, or atonic pharyngoesophageal segment.