论文部分内容阅读
本文介绍了一种日语口述数字识别机。其特点是采用了较少的样点,使用了较简单且更实用的电子线 “0” “1” “2” “3” “4” /rei/ /iti/ /ni/ /san/ /jon/ /zero/ /i/对于有两种发音的数字,如“0”,“4”,“7”,“9”,在实验中只取第一行发音。对于这些发音,经研究后,可提取八个识别特征: A.被划分的语音分段数,浊音分布; B.在第一个浊音间隔开始时,第一和第二共振峯在F_1-F_2图上的位置;
This article describes a Japanese oral digital recognition machine. It uses fewer samples and uses a simpler and more practical electronic wire. “0” “1” “2” “3” “4” / rei / / iti / / ni / / san / / jon / / zero / / i / For numbers with two utterances, such as “0”, “4”, “7” and “9”, in the experiment only the first line is pronounced. For these pronunciations, eight identification features can be extracted after the study: A. Number of divided speech segments, voiced sound distributions; B. At the beginning of the first voiced interval, the first and second formants are found at F_1-F_2 Location on the map