Physical modeling of the feedback path in hearing aids with application to adaptive feedback cancellation

Abstract

Hearing aid system modeling based on two-port network theory has been used previously to study the forward gain and the feedback path in hearing aids. The two-port modeling approach is employed in this work to develop an analytic model of the feedback path by reducing the model matrices to simplified analytic expressions. Such an analytic model can simulate the frequency response of the feedback path given the values of relatively few physical parameters such as vent dimensions. The model was extended to include variability in the feedback path due to slit leaks, for example. The analytic model was then incorporated in an adaptive feedback cancellation system, where the physical parameters of the model were adapted to match the actual feedback path and cancel the feedback signal. In the initial stage of this study, the ability of the model to match the frequency response of various measured feedback paths was studied using numerical optimization. Then, an adaptive filtering configuration based on the physical model was developed and studied using computer simulations. Results show that this new approach to adaptive feedback cancellation has the potential to improve both adaptation speed and performance robustness.