Font Size: a A A

Development Of Speech In Noise Audiometry Materials In Mandarin And Design Of A Computerized Adaptive Algorithm For Sentence Recognition Threshold Test In Babble Noise

Posted on:2013-04-12Degree:MasterType:Thesis
Country:ChinaCandidate:A T ChenFull Text:PDF
GTID:2234330362469552Subject:Biomedical engineering
Abstract/Summary:PDF Full Text Request
Speech audiometry has been widely used in audiology clinical practice andrehabilitation evaluation in China. It is useful and effective to measure speechrecognition in noise for assessing the capability of communication in patientswith noise-induced hearing loss, presbycusis, and pre-lingual hearing loss, as wellas evaluating the fitting of hearing aids. Speech recognition threshold (SRT) isdefined as the lowest speech signal level which is required to help people withhearing loss to understand50%of the speech test materials, which is also calledspeech reception threshold (SRT). SRT is used to evaluation the severity ofspeech recognition impairment.With the development of audiology in China, more and more attention hasbeen paid to Chinese mandarin speech audiometry. Meanwhile, computerizedtechnology was induced to speech audiometry. One of the greatest challenges in Chinese mandarin speech audiometry is how to approach SRT quickly and assessthe severity in communication capability.1. Because the recognition capability in noise conditions can not bereckoned with by results of pure tone audiometry, the difference inperformance in noise environment between people with hearing loss andthose with normal hearing is difficult to measure and quantify.2. Conventional speech audiometry measures the capability of speechrecognition in noise by plotting a performance-intensity function andfind the threshold SNR at which50%materials was understand. Such amethod is time consuming and tends to make patients tired, so as toaffect test results.This study was focused on the methods to measure the loss in signal-to-noise ratio and SRT in noise quickly in Chinese mandarin speech audiometry. Wehave:1. Developed a set of Quick Speech-in-Noise sentence lists for quantifyingthe capability of hearing speech in noise environment quickly, andtested the equivalence between these lists.2. Designed a computerized adaptive algorithm for sentence recognitionthreshold test in babble noise and verified its reliability in a group ofyoung men with normal hearing, obtaining the norms of sentencerecognition threshold (SNR50) using this algorithm.Innovations in this study:1. The concept of SNR loss was put forward and a quick method to obtainSNR loss was established in our study. SNR Loss is defined as the valueof dB which was needed additionally in SNR to help patients withhearing loss to understand50%of the speech in speech-in-noise test, compared with normal hearing subjects. This can be considered as thedifference between the SNR with which hearing loss patients score50%(SNR50) and that in normal hearing subjects (SNR50N).2. A computerized adaptive algorithm for sentence recognition threshold innoise was designed and tested through customized software, whichhelps to improve clinical efficiency.
Keywords/Search Tags:Speech recognition in noise, Signal-to-noise ratio, signal-to-noiseratio loss(SNR Loss), Adaptive algorithm
PDF Full Text Request
Related items