论文部分内容阅读
本文描述了一实验型语音识别系统,它采用了沃尔什分析和动态规划方法来识别不连续的、与讲话人有关联的语音。该方法的目的是利用容易买到的微处理器来研制一种廉价系统,以便根据运行速度检验算法。该系统采用 Z80微处理器,其软件在 CP/M 操作系统下开发,软件是模块化的,并且几乎全部用 Pascal语言写成,因此优化非常方便。当前计划的目标是根据本文提出的算法,提出一个能实时工作的系统的技术规范。
This article describes an experimental speech recognition system that uses Walsh analysis and dynamic programming to identify discrete, speaker-related speech. The purpose of this method is to develop an inexpensive system using readily available microprocessors to verify algorithms based on speed of operation. The system uses the Z80 microprocessor, the software developed under the CP / M operating system, the software is modular, and almost all written in Pascal language, so the optimization is very convenient. The goal of the current plan is to propose a technical specification of a system that can work in real time based on the algorithm proposed in this paper.