Britton Olson and Dr. Scott L. Thomson, Mechanical Engineering
Many voice disorders impede normal speech production. In severe cases such as laryngeal carcinoma, the total laryngectomy procedure is required, resulting in the loss of the natural ability to produce sound for speech. In this case synthetic forms of speech are employed. One method of synthetic speech that has demonstrated potential is a “voice-producing element” (VPE) . Recent VPE designs consist of two latex membranes within a cylindrical housing that is placed in the tracheoesophageal wall in a “double-reed” configuration. Air flowing from the lungs between the membranes results in regular, large-amplitude, self-sustained vibrations. The membranes are loaded on the “non-flow” side with lead-alloy weights to reduce vibration frequency as illustrated in Fig. 1
Two challenges associated with the current VPE design are related to frequency of vibration and fabrication materials. The current frequency of the VPE is around 240 Hz, compared to the average frequency of around 120 Hz for an adult male and 210 Hz for an adult female. Designs with lower frequency are therefore required for use by adult males. Further, to achieve optimal performance regarding frequency and sound production, VPE prototypes have been fabricated using materials that are not biocompatible. The present research is to investigate new VPE designs that feature lower vibration frequencies and that could be fabricated using biocompatible materials.
Three-dimensional finite element (FE) models of VPE prototypes have been previously developed and satisfactorily validated. While it is expected that biocompatible materials with properties similar to latex could be implemented, an alternative to the attachment of lead weights is necessary, considering issues of biocompatibility as well as durability (the latter to eliminate the potential for separation of the weights in vivo).
A new design researched under this ORCA grant is based on the previous double-membrane configuration, only with the lead weights replaced by a secondary flexible membrane as shown in Fig. 2. Only one-fourth of the VPE is modeled, assuming symmetry about the x-y and x-z planes. The model consists of separate but fully-coupled fluid and solid domains. The fluid domain has a constant inlet pressure of 0.9 kPa, which is comparable to that produced by the lungs during loud speech. The working fluid is air with density 1.2 kg/m3 and dynamic viscosity 1.8×10−5 Pa∙s. Other boundary conditions include zero pressure at the flow outlet, no-slip wall conditions along rigid walls, and consistent stress and velocity along the fluid-membrane interface. A contact plane is located near the x-z symmetry plane to prevent complete fluid mesh collapse (see Fig. 2).
The solid domain includes two membranes. The primary membrane has thickness 0.13 mm, modulus of elasticity 2 MPa, and density 1085 kg/m3. This membrane is the same as that which was used in earlier designs. Attached to this membrane is a secondary membrane with modulus of elasticity 500 kPa, density 1085 kg/m3, and thicknesses, w, ranging from 0.13 mm to 1.04 mm. The commercial finite element code ADINA is used for model implementation; details can be found in.
The FE model predicts regular, flow-excited oscillations; Fig. 3 shows the time history of the y-displacement of a node on the primary membrane surface for a 0.78 mm thick secondary membrane. The oscillations reach a relatively stable pattern of oscillations, and contact is evident as the membrane approaches the symmetry plane (maximum displacement).
A strong dependence of secondary membrane thickness on vibration frequency is evident in the plot shown in Fig. 4. A minimum in frequency exists around a secondary membrane thickness of 0.65 mm; this corresponds to 5 times the thickness of the primary membrane. This behavior is attributed to the counteracting effects of added mass and stiffness on the system. At low thickness the influence of increasing the mass is more significant in reducing frequency than the accompanying increase in stiffness. At higher thicknesses, the contribution of the membrane stiffness overcomes that of the added mass, resulting in an increase in frequency.
Current research includes investigating two additional aspects of the secondary membrane. The first is to examine the effect of using a material with lower modulus of elasticity values. It is expected that this will lower the overall frequency, and likely possess a similar frequency vs. membrane thickness dependence to that discussed above. The second aspect is to explore designs with multiple, separated secondary membranes, similar to the initial mass-loaded design shown in Fig. 1, although with flexible, low-modulus materials to provide the necessary reductions in frequency and achieve acceptable voice synthesis.