Parametric Evaluation Of Adaptive Feedback Control Using Variable StepSize Algorithm In Digital Hearing Aids

Abstract

The acoustic feedback signal (AFS) generated in hearing aids (HA) will emulate an ambient linkage between the loudspeaker and microphone signals, which is due to the superimposing of the loudspeaker signal (LS) on to the input sound signal. The AFS will also generate an undesired whistling sound to the HA user, during which the gain of the hearing system would get reduced. newlineFeedback cancellers have been employed for canceling the acoustic feedback, by estimating the feedback path impulse response, to produce the replica of the original AFS, which has then subtracted from the microphone output signal, to achieve a feedback eradicated signal. newlineHowever, the bias introduced into the system due to the correlation of input and output signals, have been attempted to be decorrelated using several approaches like high-frequency gain control, phase cancellation, notch filtering methods, etc. newlineIn this work, the adaptive filter, incorporated with the proposed Auto-correlation Variable step-size Least-Mean-Squares newline newline newline newline newline newline newline(AC-VSS-LMS) algorithm, has been introduced to exclude the biasing issues, by varying the step-size (and#956;) value in the filter update. newlineThe general key of using variable step-size (VSS) to the Least-Mean-Squares (LMS) algorithm, will update the filter coefficients individually and independently to each other, that resolved the contradiction between the rate of convergence and the steady-state error by shortening the transient period of the adaptive filter implemented. newlineThe algorithm has been raised to adapt the step-size by observing the system output, which varies in proportion to the magnitude of estimated filter coefficients. The performance was evaluated for achieving steady-state measures such as mean square error, Misalignment, Added Stable Gain, and Signal-to-Noise Ratio Enhancement, to optimize the convergence rate with a modest increase in computational complexity. newlineThe proposed method demonstrated theoretically, were compared with the existing techniques and the simulation results proved that the AC-