Hearing Loss Adaptivity of Machine Learning Based Compressive Sensing Speech Enhancement for Hearing Aids

Abstract

Hearing aids are needed to compensate for various auditory losses in human speech cognition. Though many techniques have for proposed for speech enhancement for hearing aids, they are tuned for a particular loss and not robust against many losses. Most existing speech enhancement techniques lack adaptivity to various noises. This work proposes a machine learning speech enhancement technique based on compressive sensing. The proposed technique adapts its sensing to obtain noised reduced speech characteristics and then amplifies it for different hearing loss. The effectiveness of proposed solution is tested against different hearing loss and the solution is found to perform well in terms subjective and objective speech quality metrics