Development of efficient filter bank structures for audiogram matching in Digital Hearing Aids

Abstract

newlineHearing aids improve the hearing perception by selectively amplifying sound signals to newlinesuite the hearing characteristics of the patient. Filter bank has a critical role in digital newlinehearing aids. It decomposes the input signal in to frequency bands which may be processed newlineindependently to best compensate for the hearing loss. The design considerations of a filter newlinebank for compact hearing aid include minimum hardware, low power consumption, low newlinedelay and higher flexibility. newlineMost of the reported methods of filter bank design for digital hearing aids in literature are newlinebased on Parks McClellan (PM) method. Transition region is not taken in to consideration newlinein this design. So, PM algorithm is not often reliable when the number of specified bands newlineis more than two as it results in very low convergence and exhibiting large undesirable newlinemagnitude response characteristics in the transition region. In hearing aid audio spectrum, newlinesignal energy is not at all zero in between bands. So, filter transition band characteristics newlineare very much important and cannot be treated as don t care region. The work presented in newlinethe thesis explores the different novel filter bank architectures based on constrained least newlinesquare and minimum phase methods, suitable for audiogram matching in digital hearing newlineaids.Audiogram provides a visual representation of hearing capacity of a person at various newlinefrequencies. Constrained least square (CLS) method is used for the filter bank design in the newlinefirst part of the work. Here the integral square error is calculated over the entire cycle of newlinefilter response including the transition band. So, signal energy in the transition band is also newlinewell taken in to consideration in the design phase.