【作者】 张立；

【导师】 李瑞峰；

【作者基本信息】 哈尔滨工业大学 ， 机械电子工程， 2006， 硕士

【摘要】 随着现代科技和计算机技术的不断发展,人们在与机器的信息交流中,需要一种更加方便、自然的方式,而语言是人类最重要、最有效的信息源。另外实现人机之间语音交互,让机器“听懂”人的话也是人们梦寐以求的事情。随着语音识别技术的出现,使得这一理想得以实现。特别是在智能服务机器人系统中,基于语音识别的人机交互是最为直接的交互手段。近年来,随着智能化小产品和消费类电子产品的迅速发展,语音识别系统大量从PC平台转移到嵌入式设备中,嵌入式语音识别技术得到越来越多的应用。本课题主要研究基于嵌入式语音识别技术的智能家用服务机器人人机交互控制系统,用户可通过语音命令实现对服务机器人的运动控制,另外还可实现用户与机器人之间的语音交互。本文在介绍了语音识别基础知识及常用语音识别算法的基础上,根据嵌入式系统的特点和课题要实现的具体功能,给出了系统硬件和软件实现的总体方案。以台湾凌阳公司SPCE061A语音芯片为该嵌入式系统的核心芯片,设计出家用服务机器人嵌入式语音识别交互硬件系统。包括:芯片及其外围电路、电源电路、语音输入电路、音频输出电路、系统调试电路、串口通讯电路等。采用计算量相对较小和对系统资源要求较少的动态时间规整(DTW)算法实现语音识别功能;进行语音命令训练存储、语音交互、交互所需语音资源的存储调用、串口通讯等软件设计,实现人对机器人的语音控制和人机语音交互功能。为了提高系统的语音接收和识别能力,进行多通道语音输入电路设计,分析了不同位置状态下多通道语音接收与识别能力。最后对嵌入式语音识别交互系统进行试验验证,测试了不同条件下的语音识别率和识别稳定性,分析了异性间识别率差异的原因,测试了多通道语音输入电路在不同状态下的性能。最后对服务机器人系统进行了语音控制试验和人机交互功能测试。更多还原

【Abstract】 With the development of modern science and computer technology, people need a more advantageous and natural method to communicate with the machinery. In terms of this, language is the most important and effective information source. And further more, it has always been a dream for human being to realize man-machine interaction via speech. Thanks to the emergence of speech technology, this dream comes true. Especially in the intelligent service robot system, it is the most immediately measure to realize speech communication.Presently, with the swift development of intellectualized product and consumptive electron product, Speech Recognition System has transferred largely from PC system to embedded equipment, Embedded Speech Recognition System has found a wider and wider application. This paper focuses mainly on the man-machine interaction system of domestic service robot based on embedded speech recognition technology, the user can use speech to control the movement of the service robot and can talk with the robot.This paper presents the fundamental knowledge of speech recognition and several speech recognition algorithms, then we give the hardware and software design of the system according to the characteristics of the embedded system and the particular requirements of the project.SPCE061A Speech Chip produced in Taiwan is selected as the core chip in the Embedded Speech Recognition and Interaction hardware system. It mainly contains chip circuit and the periphery circuit of it, power circuit, speech input circuit, speech output circuit, debugging circuit and RS232 circuit.In the software part, Dynamic Time Warping (DTW) that needs lesser account and lesser system resources is selected to achieve speech recognition. The software of speech training, speech interaction, storage and transferring of the speech resource and RS232 communication is designed to realize the speech control and the speech interaction.The multicenter speech input circuit is designed to improve the capacity of更多还原