| 口令识别系统 |
| |
| 引用本文： | 刘承玺.口令识别系统[J].天津大学学报(自然科学与工程技术版),1989(4). |
| |
| 作者姓名： | 刘承玺 |
| |
| 作者单位： | 天津大学电子工程系 |
| |
| 摘 要： | 介绍一个计算以TMS32020信号处理器为主,由IBM—PC/XT机控制的定人口令实时识别系统。算法原理是由线性预测编码(LPC)系数转换为倒频谱(CEP)参数;在模式识别中,采用两级匹配法。由于算法的优化和高速处理芯片(TMS32020)的运用,识别率和识别速度均较高,实验结果是:10个数字(0—9)的识别率约为99.5%;50单词的识别率约为98.5%:识别时间小于0.5s。
|
| 关 键 词： | 口令 倒频谱参数 两级匹配法 |
RECOGNITION SYSTEM OF SPOKEN COMMANDS |
| |
| Liu Chengxi.RECOGNITION SYSTEM OF SPOKEN COMMANDS[J].Journal of Tianjin University(Science and Technology),1989(4). |
| |
| Authors: | Liu Chengxi |
| |
| Institution: | Department of Electronic Engineering |
| |
| Abstract: | In this paper, a real-time recognition system of spoken commands for a particular speaker is introduced. The computation of the system mainly depends on signal processor TMS32020 and the system is controlled by IBM-PC/XT. The algorithm is to transform linear predictive coding (LPC) cofficients into parameters of the cepstrum. In the pattern recognition, two-level-matching method is used. Because of the algorithm optimization and application of high-speed processor, recognition rate and speed both are higher. The experimental results are as follows: Recognition rate of ten digits (0-9) is about 99.5%; Recognition rate of fifty Chinese characters is about 98.5%; Recognition time is less than 0.5 second. |
| |
| Keywords: | spoken command parameters of cepstrum two-level-matching method |
| 本文献已被 CNKI 等数据库收录！ |
