A Decision-Directed Adaptive Gain Equalizer for Assistive Hearing Instruments

Abstract

Assistive hearing instruments have a significant impact on speech enhancement when the signal-to-noise ratio is low. These instruments are usually developed using the conventional adaptive gain equalizer (AGE), which has low computational complexity and low distortion in real-time speech enhancement. The conventional AGEs are intended to boost the speech segments of speech signals but they are incapable of suppressing noise segments. The overall speech quality of the assistive hearing instruments may be reduced, as the noise segments still cannot be filtered out. In this paper, a decisiondirected AGE is proposed for assistive hearing instruments. It aims to overcome the limitation of the conventional AGE, which is capable only of boosting speech segments in noisy speech but incapable of suppressing noise segments. The proposed approach simultaneously boosts the speech segments and suppresses noise segments in noisy speech. Experimental results with different types of real-world noise indicate that the proposed method achieves better speech quality than does the conventional AGE. The resulting method provides an improved functionality for assistive hearing instruments.Index Terms-Adaptive gain equalizer (AGE), assistive hearing instruments, heuristic algorithms, particle swarm optimization (PSO), perceptual evaluation of speech quality (PESQ), single channel filter, speech booster, speech enhancement.

Hyper-parameterization of sparse reconstruction for speech enhancement

“…Clearly, delivering a clean speech signal in the communication system is an important aspect of the product requirement. The objective of speech enhancement is to estimate the desired speech signal from the noisy observation, which consists of both speech and noise signals [1,2].…”