Feedbackcancellation in Vented Hearing Aids

The application of adaptive lters in feedback cancellation in vented and unvented hearing aids (HA), together with their feedback path properties when worn by a head and torso simulator, have been analyzed and studied in this project. A vented and an unvented in the ear (ITE) HA were mounted in a head and torso simulator (HATS) and the feedback paths were measured with different objects placed near the HAs. Adaptive feedback cancellation (AFC) algorithms based on the least mean square(LMS) algorithm have been simulated with white noise and speech input signals. Due to the short distance between the microphone and receiver in HAs, the feedback and input signal can be correlated in the case of a highly autocorrelated input signal such as speech. The simulations performed show that this correlation does not make the estimation of the feedback path accurate enough, and hence the feedback cancellation not effective. Decorrelation methods were studied and the delay in the forward path option was simulated. The simulation results show that this is an effective solution to decorrelate the input and feedback signals and hence achieve a better estimation. When a white noise input signal was used, an added stable gain(ASG) of 20 dB was obtained in the simulations. The LMS approach was implemented in a digital signal processor as an adaptive feedback canceller. For a white noise input signal the ASG of the AFC was found to be 12dB.