실시간 MBSS와 SVM을 이용한 보청기 잡음제거 알고리즘 개발

초록

This study suggests an adaptive noise reduction algorithm which combines an environment classification system and multi-band spectral subtraction (MBSS). Environmental conditions were divided into four classes such as babble, car, traffic and white noise. Support vector machine (SVM) using a radial basis function kernel was used to classify the environmental conditions. The classification accuracy was as high as 97%. Depending on the environmental conditions, additional weighting factors for nine frequency bands were determined and applied to MBSS