A study on howling canceller using quasi-whitened input signals

Acoustic feedback called howling of hearing aids causes annoying effect for users. The adaptive filter is used to model the acoustic feedback path and cancel its effect. In this article, firstly, conventional methods, the plain LMS method and the modified LMS method are briefly reviewed. And then, a problem of the conventional methods, namely, falling in biased estimation, is described. This is due to: (1) correlation between the input signal of the adaptive algorithm and the observation noise; (2) uneven frequency allocation for FIR type notch filter. In order to cope with the problems, a howling canceller with IIR notch filters, in which the input signal of the adaptive algorithm is quasi-whitened, is proposed. Finally, from extensive simulations, the validity of the proposed method is verified.