Computationally efficient active noise reduction in headsets

In order to improve the passive attenuation of hearing protection headsets, active noise reduction (ANR) techniques are usually applied. These ANR-techniques accomplish the active attenuation of the disturbing acoustical noise using an out-of-phase antinoise. The antinoise destructively interferes with the disturbing noise close to the humans ear drum. The generation of the antinoise can be conducted using different control strategies. However, due to variable system plants, often adaptive control strategies are chosen. Even though such adaptive systems effectively attenuate the disturbing noise in a wide frequency range, a major disadvantage is the computational effort linked to the large amount of controller parameters. The controller parameters have tobe updated by the adaptive algorithm and the resulting computational effort makes the application of expensive digital signal processors unavoidable. For this reason, no commercial products realizing adaptive broadband techniques are on the market yet. In this paper, a partially-adaptive control approach is introduced which permits the reduction of the computational effort in comparison to conventional and fully adaptive ANR-controllers. The noise reduction performance as well as the computational efficiency of the proposed control strategy is presented.