Hearing aids (HAs) help people who are hard of hearing to improve speech perception in scenarios such as, speech in background noise, when hearing from a distance or when communicating with speakers with soft voices. Recently, to improve the potential noise reduction performance in HAs, methods that include remote microphones (RMs) to sample a wider spatial field have been proposed in literature. The signal transmitted by the RM have commonly been incorporated as an extra channel by the HA beamformers. Moreover, they incorrectly assume that the RM signal is transmitted instantaneously to the HA unit, in contrast to real world applications. The remote and local HA microphones are located on separate devices and their signals need not be synchronously sampled. Moreover, there will be a time difference of arrival (TDOA) between the acoustic signal received by the local HA microphones and the wirelessly transmitted signal received by the HA unit from the RM, which varies depending on the distance between the target source and the microphones, the distance between the microphones and the wireless transmission protocol used. As the TDOA increases, the potential benefit of using the RM reduces, thereby rendering them nearly useless. We use the binary estimator selection (BES) strategy that considers the fact there exists a TDOA between the HA and RM signals. We use linear minimum mean-square error (MMSE) filters, to estimate the desired target signal, using the HA and RM signals independently, and use the BES strategy to select between the two estimates in each time-frequency tile based on their corresponding normalized mean-square errors (nMSEs). By doing so, the proposed BES method, picks the estimate with the lower nMSE in each time-frequency tile. The benefit provided by this strategy is to use the RM signal more, when the TDOA is low, and to use the HA signals more when the TDOA is large. Therefore, we use the benefit of both the HA and RM signals, particularly when there is a TDOA between the signals received at the HA unit.

Funding: This project has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No. 956369.