台灣地區噪音下漢語語音聽辨測驗 (Taiwan MHINT) 主要在評估受試者的百分之五十語句聽辨閾值。測試語料共有十二個測驗語句表與四個練習語句表，一個語句表包含二十個句子，每個語句由十個中文字組成。語句接收閾值分別在三種噪音環境以及安靜環境下進行測量，噪音強度固定在65 dB的音量，並以適性方式來改變語音訊號強度。本研究目標為發展台灣地區噪音下漢語語音聽辨測試之軟體，並檢驗其與HINT Pro市售儀器有同等測驗難度與效果。 本研究對9名聽力正常和2名聽力受損的個案透過耳機聆聽方式進行噪音下語音聽辨測驗，個案皆為智力正常的國立中央大學學生以及研究生，年紀在20歲至27歲之間，隨機先後進行市售儀器測驗與研究發展軟體測驗，並蒐集測驗結果進行比較。結果表現本研究發展軟體在四種環境下的語句接收閾值與市售儀器無顯著差異，顯示本研究軟體與市售儀器有同等測驗難度與效果，並且對於聽力受損的個案本研究發展軟體與市售儀器也有同等的鑑別度。此外，相隔兩個禮拜後再重新測驗，重測的測驗結果與第一次的測量結果無顯著差異，表示本研究發展軟體具有重測可靠度與對同一位受測者測量結果有再現性。 ;Taiwan Mandarin Hearing In Noise Test (Taiwan MHINT) was developed to the 50% evaluate reception threshold of sentences (RTSs) of listeners. The test speech materials involved twelve test sentence lists and four practice sentence lists. Each list consisted of twenty sentences and each sentence consisted of ten Mandarin characters. The reception thresholds of sentences of listeners were measured in three different noise conditions and the quiet condition. The noise level was fixed at 65 dB and the speech level was adjusted in an adaptive procedure. The objective of this study was to develop the software for Taiwan Mandarin Hearing In Noise Test, and to examine the test difficulty and effectiveness of the proposed software and the HINT Pro commercial instrument. In this study nine normal-hearing subjects and two hearing-impaired subjects were involved. All the subjects aged between 20 to 27 were with normal intelligence and were graduate and ungraduate students of the National Central University. The subjects participated in the hearing in noise test, which was carried out in a random test order with the commercial instrument and the proposed software by using earphone. The results showed that there was no significant difference between the RTSs of the commercial instrument and the RTSs of the propose software test in four different noise and quiet conditions. Hence the commercial instrument and our software have equivalent test difficulty and effectiveness as well as similar degree of discrimination for hearing-impaired listeners. Furthermore, a second test was repeated after two weeks and there was no significant difference between the repeated test results and the first test results. It is evident that the developed software of this research has the reliability of retesting and the reproducibility of measurement results for the same subject.
