Open AccessEfficient hearing aid algorithm using DCT with uniformly re-sampled and recursively modified audiogram values
Author(s) -
K. Ayyappa Swamy,
Zachariah C. Alex
Publication year - 2021
Publication title -
indonesian journal of electrical engineering and computer science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.241
H-Index - 17
eISSN - 2502-4760
pISSN - 2502-4752
DOI - 10.11591/ijeecs.v23.i1.pp237-246
Subject(s) - audiogram , precomputation , discrete cosine transform , hearing aid , algorithm , computer science , speech recognition , filter bank , filter (signal processing) , mathematics , hearing loss , audiology , medicine , artificial intelligence , computer vision , computation , image (mathematics)
People with the hearing problems have different listening preferences and characteristics in hearing loss. So, hearing aids need algorithms that provide amplification based on frequency, so that the hearing-impaired persons can use hearing aids comfortably for a long duration. In this paper, a new algorithm is proposed for hearing aids in order to compensate for sensorineural and conductive hearing loss using discrete cosine transform (DCT). DCT coefficients of the input audio signal are multiplied with uniformly resampled and recursively modified audiogram values to compensate for hearing loss. This algorithm comprised of 4 stages namely precomputation to calculate gain values from audiogram, DCT, gain adjustment, and inverse DCT. In the above stated stages except precomputation, each stage requires only one matrix multiplication, which makes the proposed algorithm computational efficient. Performance of the proposed algorithm is compared with uniform filter banks, non-uniform filter banks, variable filter bank and reconfigurable filter banks. The algorithm is tested using audiograms with four different hearing loss cases. It is proved that the proposed algorithm provides less complexity, minimized delay and better matching with all types of audiograms, further, it also avoids degradation of audio signal due to sampling rate conversions in variable and reconfigurable filter banks.