Research System Of Speech Quality Evaluation For Hearing Aids

Hearing aid is an important and assistive device to compensate hearing for the hearing disabled, proceed with language rehabilitation training and contact with others normally. The quality of hearing aid is related to the hearing impaired physical and mental health directly. With the developing of digital and signal technology, hearing aids have lots of high-end performances. The quality of speech is one of the significant performances. At present, there is not much research for the speech quality evaluation of the hearing aids. PESQ algorithm is based on the perceptual model and has the highest correlation degree with subjective MOS score. This paper introduces PESQ algorithm in detail and improves part of modules. Design filters to simulate the hearing losing curves. Base on MATLAB language and Visual Basic language, the paper designs a system for evaluation the hearing aids speech quality. Depend on the experiments, PESQ and improved algorithm can evaluate different hearing aids. The improved algorithm increases working efficiency.The mainly work of this paper is including:1) The research development status of two evaluation methods are introduced, it includes subjective evaluation and objective evaluation. This paper chooses PESQ algorithm as the focus research. A scheme of evaluation system is proposed. The following researches the modules of speech database, playing speech, test system, filters and speech quality evaluation.2) Basis for the typical hearing losing curves in IEC 60118-15, the topic selects the least square method to design FIR filters. The filters simulate the hearing losing curves N1-N4. Pure signals and speech signals pass through the filters to Analyze filtering effect of designed filters.3) Study the principle and realization process of PESQ. According to this algorithm used to evaluate the hearing aids speech quality and aural masking in human ears. The topic gives partly improvement such as Intermediate Reference System (IRS) filter, the loudness transform, aggregation of the disturbance.4) In the test platform, the experiment chooses two types of different hearing aids which are downloaded different hearing curves. Analyze and summary PESQ and improved methods from evaluation score, time complexity and so on.