论文部分内容阅读
语音识别技术越来越广泛的应用于嵌入式系统平台.在分析基于HMM语音识别原理的基础上,设计并实现了一个基于ARM9和嵌入式Linux系统的特定人、孤立词和小词汇量的嵌入式语音识别系统.采用S3C2410微处理器和嵌入式Linux操作系统,将交叉编译后的语音识别C语言程序移植入嵌入式Linux操作系统的文件系统,实现语音识别系统的功能.给出了系统整体的软硬件框架,并给出了实时输入识别0到9的汉语数字语音识别的实验结果,探讨了在不同码本长度和不同隐含马尔可夫模型状态数的识别性能,为语音识别技术在嵌入式设备实际应用提供参考.“,”Speech recognition technology increasingly extensive used in embedded system platform.Designed and implemented a embedded speech recognition system which base on ARM9 and embedded Linux system of the specific,isolated words and small vocabulary.Used S3C2410 microprocessors and Linux operating systems,Transplant the C language of speech recognition program to the embedded Linux operating system's file system when had been cross-compiled. Gave the system's hardware and software framework,and gave the experiments results which are real-time recognition of the 0-9 Chinese figure speech,explore the recognition performance when in the different code book length and different state numbers of the Hidden Markov Model, provision the speech recognition technology for embedded devices in the practical application's reference.