CAJ | 학술논문

从英语口语网络机考的特点出发,基于计算机技术的角度,结合并利用对于传统的谱相减降噪声技术和倒谱均值规整降噪技术的修改以及等同于连续状态隐马尔可夫模型的概率尺度DP算法,提出了一种在英语口语测试系统中非特定人的网络语音识别方案,即通过实际环境下的英语口语测试实验,验证了新方法的有效性;从基于英语口语测试场地改建的角度,阐述了一些除噪的措施和建议,试图设计出一种解决噪声的完美方案,能够让考生在完美＂静音＂空间里进行英语口语网络机考。
종영어구어망락궤고적특점출발,기우계산궤기술적각도,결합병이용대우전통적보상감강조성기술화도보균치규정강조기술적수개이급등동우련속상태은마이가부모형적개솔척도DP산법,제출료일충재영어구어측시계통중비특정인적망락어음식별방안,즉통과실제배경하적영어구어측시실험,험증료신방법적유효성;종기우영어구어측시장지개건적각도,천술료일사제조적조시화건의,시도설계출일충해결조성적완미방안,능구양고생재완미＂정음＂공간리진행영어구어망락궤고。
By analyzing the characteristics of intemet-based English oral test 0B-EOT）, from computer-assisted technology in the first place, we present a method of disposing noise utilizing DSP in real environment recognizing English speech. Through experiments of iB-English oral speech recognition in the noisy outdoor environment, the effectiveness of the method is confirmed. From the construction in space environment of iB-EOT, We propose some suggestions and measures to get rid of noise. Through these basic methods, we intend to find an ideal way to solve the noise problem in iB-EOT so that test-takers can do their performance without any noise.