Perceptually motivated binaural beamforming with cues preservation for hearing aids
Hala As'ad, Martin Bouchard, Homayoun Kamkar-Parsi · 2016
In this contribution, a new binaural beamforming algorithm for hearing aids is introduced in order to generate binaural outputs with a good trade-off between noise reduction and binaural noise cues preservation. This includes cues preservation for the directional noise sources as well as diffuse-like background noise. The new algorithm is based on selection and mixing processes in the Time-Frequency domain (T-F) and is applicable for realistic T-F decompositions. The selection and mixing are done with a classification algorithm based on the complex coherence between the standard beamformer outputs and other available signals. To validate the performance of the proposed algorithm, simulations have been performed using recorded signals provided by a hearing aid manufacturer. The results indicate that the proposed algorithm can provide a good trade-off between noise reduction and binaural noise cues preservation, even for complex acoustic environments.