AUTOMATIC SPEECH RECOGNITION BY ZERO-CROSSING INFORMATION.

被引:0
|
作者
Shiraishi, Mikio [1 ]
Watanabe, Kajiro [1 ]
Tanaka, Tomoharu [1 ]
机构
[1] Hosei Univ, Dep of Electrical, Engineering, Tokyo, Jpn, Hosei Univ, Dep of Electrical Engineering, Tokyo, Jpn
关键词
INFORMATION SCIENCE - Computer Applications;
D O I
暂无
中图分类号
学科分类号
摘要
A system for speaker-dependent isolated word recognition is described. The system consists of a single-board preprocessing hardware and a microcomputer system. In this hardware, speech signal is divided into two frequency bands, and zero crossings of speech signal in the respective frequency bands are counted. Recognition performance of the system was investigated. The recognition rate for a set of Japanese numeral word (0 to 9) which was uttered by 5 male speakers was 96. 6%. DP matching implementation was used in this experiment.
引用
收藏
页码:95 / 106
相关论文
共 50 条