计算机语声命令输入系统  

A Micro-Computer Based Speech Command Input System

作　　者：陈锡先[1] 蔡长年[1] 郭鹏 

机构地区：[1]北京邮电学院电信工程系

出　　处：《北京邮电学院学报》1991年第3期60-66,共7页

摘　　要：本文介绍的计算机语声命令输入系统能够通过话筒直接输入语声命令,同时也可用键盘进行操作.系统采用经过实时优化的连续矢量量化聚(CVQC)算法,可在以 TMS32010为主体的高速信号处理板(UDSP-10)上实时进行语声命令的训练和识别。识别响应时间小于20ms.本文详细讨论了系统的硬件结构,识别算法和系统接口软件。最后给出了系统识别率指标。This paper introduces a micro-computer based speech command input system which enables the computer to accept the spoken voice commands through the microphone beside the input from the keyboard.A continuous VQ clustering (CVQC)based training and pattern matching algorithm is optimized for real-time implementation in a TMS32010 based high-speed Universal DSP board(UDSP-10). The maximum response time is less than 20 ms for 40-word templates.A detailed analysis is given to the hardware structure,recognition algorithm,and the interfacing software.Finally,the recognition results for a number of commonly used vocabularies are also presented.

关 键 词：计算机 输入系统 语言识别 声控 

分 类 号：TP334[自动化与计算机技术—计算机系统结构]