Audio Classification And Scene Recognition and for Hearing Aids

Abstract

Modern hearing aids provide a variety of options and settings that can be tuned based on the audio environment. We provide a robust audio classification system that could be used to switch automatically between different hearing aid algorithms based on the auditory scene. The classification system uses features derived from a model of the auditory system in conjunction with an AdaBoost-based classifier. We show that our system achieves high accuracy and good generalization with only one second segments. We also demonstrate the auditory scene recognition capabilities of the classification system.