Improved design of microphone-array hearing aids

Abstract : A common complaint of hearing aid users is the difficulty encountered when listening to a talker in a noisy environment. Conventional hearing aids amplify all sounds without discriminating between the desired source (target) and background noises (jammers). These devices increase the overall sound levels, but do nothing to improve target-to-jammer ratio (TJR). Research on microphone array hearing aids is motivated by the lack of success of single-microphone systems, as well as the documented advantages of binaural hearing and multiple-element sensing systems. Array processing can be classified as either fixed (time invariant) or adaptive (time varying). Previous work on microphone array hearing aids has demonstrated that under certain conditions, adaptive arrays can provide significantly better performance than simpler fixed arrays. The benefit of adaptive systems is realized when the input TJR is low and when the signals arriving via direct paths are stronger than the reflections. This benefit is reduced or eliminated at high TJR or in strong reverberation. This work studies modified adaptive algorithms to improve performance at high TJR and in reverberation; it also provides complete specifications for the design of an adaptive microphone array hearing aid. In particular, two previously proposed ad hoc methods for controlling adaptation at high TJR are analyzed and evaluated. The results confirm the usefulness of these methods and provide gnidelines for selecting relevant parameters in anechoic and reverberant environments. In addition, an analysis of the specific causes of target cancellation in reverberation reveals that a simple set of parameter choices can solve this problem. Computer simulations of the complete system demonstrate its benefits in a variety of acoustic environments. Steady-state results show that the system provides very large improvements i