In modern hearing assistant devices, it is common practice to implement adaptive beamforming. Microphone array adaptive beamforming applied to hearing aids must improve the intelligibility of a target sound source, while preserving natural spatial cues for the user. As discussed in, Binaural MVDR Beamforming, is able to enhance the target signal and preserve the spatial cues of the desired talker, but the auditory cues of the interfering sources are lost as part of this process.
The Binaural MVDR beamformer is designed such that interaural transfer function (ITF) of the input and output should be equal:

$\min_{w}{w}^H{S}_{xx}w$ subject to ${w^h}d =1$

${oITF}_{d} = {({w_L^H}d) / ({w_R^H}}d) = {iITF}_{d}$

In order to preserve the cues of the interfering sources, additional constraints are introduced into the system, turning the Binaural MVDR into the Binaural Linearly Constrained Minimum Variance (LCMV) beamformer. Additional constraints are made for each noise source, such that:

${oITF}_{n} = {({w_L^H}h_n) / ({w_R^H}}h_n) = {iITF}_{n}$

Given M microphones placed at each ear, there are 2M degrees of freedom (DOF) in the design. 2 degrees are freedom are used for the minimum distortion for the desired signal in each ear (${w^h}d =1$). Leaving 2M-2 DOF for spatial cue preservation, and the remaining DOF for noise reduction. Therefore, there is a tradeoff between noise reduction and auditory cues. Due to variety of potential acoustic environments and the hearing characteristics of the patient, this tradeoff should be made controllable for the patient.

VOCAL Technologies offers custom designed solutions for beamforming with a robust voice activity detector, acoustic echo cancellation and noise suppression. Our custom implementations of such systems are meant to deliver optimum performance for your specific beamforming task.