Research On Adaptive Acoustic Feedback Cancellation In Hearing Aids

Digital hearing aids （DHA） could help people with hearing loss improve auralcomprehension greatly. The acoustic feedback occurs due to the acoustical coupling betweenthe loudspeaker and the microphone in combination with the usually high hearing aidamplification. Acoustic feedback limits the maximum stable gain of DHA and deterioratesspeech intelligibility. This paper research focused on adaptive acoustic feedback suppressionbased on proportion adaptive filtering algorithm, and proposed a kind of variable step algorithmto enhance feedback cancellation. The main work can be summaries as follows:1、This paper researches the principle of audition handicap, and analyzed the auditioncharacteristic of people with hearing loss. Several technologies used in hearing aids arediscussed, including compression amplification, speech enhancement, frequency transposition,source location and some conventional methods of acoustic feedback suppression.2、 Acoustic feedback cancellation in digital hearing aids bases system identification,sothis article researches the theory of system identification by adaptive filter deeply. Two kinds ofadaptive filtering algorithms are researched：Least Mean Square and Recursive Least Square,and the advantage and disadvantage are considered. Normalized least mean square is deducedin detail which is widely applied to acoustic feedback cancellation.3、Since the feedback path is sparse impulse response，this paper applies IPNLMS algorithm（Improved Proportion NLMS） to feedback cancellation to improve convergence speed thanNLMS considerably. The research shows that the requirements of fast convergence and lowsteady-state misalignment are conflict. This paper proposes a novel variable step-size algorithmto solve the contradiction. The algorithm regulates step size in accordance with filter weightgradient. A decrease in the gradient of the filter weights causes the step size decrease. Thealgorithm is insensitive to noise and suitable for digital hearing aids acoustic feedbackcancellation. The simulations result shows that proposed algorithm achieves excellentperformance than NLMS and IPNLMS in feedback cancellation system.