Audio feature clustering for hearing aid systems
Nasim Shams, Behnaz Ghoraani, Sridhar Krishnan · 2009
This paper presents a novel approach for classification of audio signals for a noise free hearing aid system. Due to the large number of people who suffer from hearing problems, the impact of developing such a system is significant. Using the proposed classification tool, we are able to discriminate the environmental noise from speech, and then prevent the noise signals from being magnified by the hearing aid. A set of features is extracted in the time-frequency domain due to the non-stationary nature of the signals. Then a novel classification method is applied, which is based on the self organizing tree maps clustering algorithm followed by a fuzzy labeling of the clusters. The result show an accuracy of 96% for separating the human voice from the environmental noise.