Binaural Noise Reduction in Hearing Aid Simulations Using a Multichannel Wiener Filter
Niccolo Gans, Franz de Leon · 2019
In audio signal processing, unwanted noise is a very common and serious problem. For people with hearing loss, this problem is magnified. In this project, a two-stage binaural noise reduction algorithm using a Wiener filter is highlighted, analyzed, and viewed as a potential solution to the problem of noise reduction. Compared to other algorithms, this one is very robust and capable at enhancing noisy input signals while still preserving the natural sound spatial information within them. This work implements the algorithm and tests its efficiency in several ways, on simulations involving a listener wearing earpieces in a noisy environment. Evaluation results show that the noise in the input signals is significantly reduced, while the sound spatial information and speech intelligibility is retained to an acceptable degree. Such an effective algorithm could be useful for hearing aid technologies, as well as for future studies and work on audio signal enhancement.