CAJ | 학술논문

在分析普通话水平测试中回读、漏读和增读错误的特点的基础上，提出针对语音注音符号匹配的比较测度，统计出了常见普通话读音错误表。为适应语音注音符号的匹配，对传统的动态时间规正算法进行改造。根据普通话的朗读特点，提出回读、漏读和增读错误检测的总体算法。通过对在真实的普通话水平测试现场采集的100人的朗读语音材料进行实验，结果表明：检测正确率达到99．48％。
재분석보통화수평측시중회독、루독화증독착오적특점적기출상，제출침대어음주음부호필배적비교측도，통계출료상견보통화독음착오표。위괄응어음주음부호적필배，대전통적동태시간규정산법진행개조。근거보통화적랑독특점，제출회독、루독화증독착오검측적총체산법。통과대재진실적보통화수평측시현장채집적100인적랑독어음재료진행실험，결과표명：검측정학솔체도99．48％。
Based on analysing the errors feature in regard to back reading,missed reading and inserted reading in Putonghua proficiency test,we present the comparative measure for speech and phonetic notation symbol matching and summarise the table of common pronunciation errors in Putonghua.For adapting speech and phonetic notation symbol matching,we reform the traditional dynamic time warping algorithm. According to Putonghua reading features,we present a general algorithm of errors detection for back reading,missed reading and inserted reading.Through the experiment conducted on the voice materials of 100 persons’reading on-site recorded at the real Putonghua proficiency test,the result shows that the correct detection rate reaches 99.48%.