论文部分内容阅读
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 article describes the computer voice command input system can be directly through the microphone input voice commands, but also can be operated using the keyboard. The system uses real-time optimization of continuous vector quantization poly (CVQC) algorithm can TMS32010 as the main body of high-speed signal processing Board (UDSP-10) real-time voice command training and identification. Recognition response time is less than 20ms.This paper discusses in detail the system hardware architecture, identification algorithms and system interface software. Finally, the system identification rate index is given.