目的 獲得嘈雜語噪聲背景下漢語短句識別率-信噪比（performance-intensity, p-i）函數的斜率和50%的得分所對應的信噪比(記為snr50)，并對所有短句的同質性進行評估和調整。方法 采用16張（每張20句）新編嘈雜語噪聲下漢語普通話短句識別表作為測試材料。選取年齡為21 ～ 25歲之間、聽力/言語發育正常且日常以漢語普通話為交流方式的48名受試者按“隨機區組設計”在-1、 -4、 -7、 -10 db四種信噪比條件下進行言語識別率的心理聲學測量。使用statistica7.0軟件進行p-i函數擬合和統計分析。結果 320句語句在嘈雜語噪聲下的言語識別p-i函數曲線的閾值呈正態分布，斜率呈不規則分布。剔除p-i函數強健性增長（斜率 > 55%/db）的語句，并保留閾值變異度在 ± 2?滓內的語句，精選出同質性良好的240句。結論 逐個調整每句的snr以實現同質性，可為編制“等言語識別閾級”的句表奠定扎實的心理聲學基礎。
【關鍵詞】 言語測聽； 嘈雜語噪聲； 語句； 言語識別閾； 同質性
homogeneity of mandarin sentence intelligibility in babble noise
xi xin, zhao yang, ji fei, teresa ching, han dong-yi
department of otolaryngology-head and neck surgery, chinese pla general hospital, beijing 100853, china
national acoustic laboratories, 126 greville street, nsw 2067, australia
【abstract】 objective to obtain the slope and srt of the p-i function for mandarin short sentences in 4-talker babble noise, and evaluate/adjust the homogeneity all sentences individually. methods forty-eight mandarin-speaking people aged at 18 ~ 25 with normal aural/oral communicational abilities were participated in the investigation. psychometric measures on mandarin speech intelligibility in babble noise were implemented in -1, -4, -7, -10 db snr through 16 lists (20 sentences each) of mandarin short sentences by a latin-square experimental design. estimation and analysis of p-i functions of all sentences were done by statistica 7.0. results histograms about the thresholds and slopes of p-i functions showed that the snr50 thresholds were in guassian distribution(?滋 ± ?滓 = -6.27 ± 2.04 db), but the slopes not. two hundred forty sentences with high homogeneity were chosen according to two criteria: threshold variation was within 2?滓; slope below 55%/db. conclusion homogeneity achieved by adjustment of snr for individual sentence based on psychometric measure would play an important role in development of speech test materials.......