维吾尔语语音识别技术在智能家居中的研究与应用

发布时间：2018-01-05 13:15

本文关键词：维吾尔语语音识别技术在智能家居中的研究与应用　出处：《新疆大学》2017年硕士论文　论文类型：学位论文

【摘要】：传统的电器,传统的家用设备都已经显得过期了。所以如果有一种办法使得以前的老电器不换的情况下通过智能手机实现维吾尔语语音控制控制是具备研究意义与实践意义的一个课题。处于这样的想法我开始本轮中的研究工作。智能家居系统是以家庭为单位,运用模式识别,控制指令映射,通信网络组建,语音识别,语义理解,第三方智能中控组成。目前在市场上,没有任何维吾尔语语音控制的智能家居产品。维吾尔语语音识别技术近年来逐渐成熟,新疆大学为首的一批研究者们的长期努力下新疆大学积累了相关领域的关键技术。在维吾尔语语音识别领域研究工作中新疆大学信息科学与工程学院的科研能力很强,具备长年积累。本人在研究生阶段参与了实验室的语音识别相关的项目,从中受益很多。我用三年学习时间将语音识别知识为背景结合了智能家居硬件设备,将维吾尔语语音识别技术应用到智能家居领域中。通过语音技术来识别技术与智能家居技术的整合后进行旧家电的维吾尔语语音控制,建立了方便、智能的家居控制客户端应用程序,使得系统更加智能化、人性化。本文着重研究基于CMU Sphinx 4.0维吾尔语语音识别和通过智能家居硬件接口实现语音控制电器设备(电视,灯泡,热水壶,热水器等通过开关进行控制的电器设备与通过遥控控制的电器设备)。维吾尔语短语语音识别过程由四个部分(语音信号特征参数提取、构建声学模型、构建语言模型和解码器部分的模式匹配)组成。本文对这四个部分相关的技术进行了详细的比较和分析,选择MFCC、HMM和N-gram等技术及其相关算法进行了研究,除此之外本论文中详细阐述了MFCC特征参数提取方法、HMM声学模型及其相关算法以及N-gram语言模型及平滑处理方法。所实现的语言模型与声学模型在Android系统的智能手机中实现集成到手机应用程序实现了语音控制智能家居系统中的语音识别模块。最终语音识别模块通过与手机wifi无线通讯协议相连接的智能家居硬件的API实现电器设备的控制。实现了维吾尔语语音控制智能家居的应用程序。
[Abstract]:The traditional household appliances, the traditional equipment is already expired. So if the implementation of Uyghur speech control is a topic with research significance and practical significance of the intelligent mobile phone through the old appliances before there is a way that does not change the situation. In this idea I began to research work in the current round of intelligence. Home Furnishing system is based on the family unit, the use of pattern recognition, control instruction mapping, network communication, speech recognition, semantic understanding, the third party intelligent control. At present in the market, there is no Uighur speech control of intelligent products. Home Furnishing Uyghur speech recognition technology in recent years gradually mature, a number of long-term efforts research led by the Xinjiang University of Xinjiang University under the accumulation of key technology related field. Research work in the field of Uyghur speech recognition in Information Science Xinjiang University Science and Engineering College of scientific research ability is very strong, has accumulated for many years. I participated in the speech recognition project related laboratory at the graduate level, benefit a lot from it. I spent three years learning time speech recognition knowledge as the background combined with Home Furnishing intelligent hardware devices, the Uighur language speech recognition technology applied to intelligent in the field of control. Home Furnishing Uyghur speech recognition technology and the integration of intelligent speech technology by Home Furnishing technology after the old electrical appliances, built a convenient, Home Furnishing control client application intelligence, which makes the system more intelligent and humane. This paper focuses on the research of CMU Sphinx 4 Uyghur speech recognition and intelligent Home Furnishing through hardware interface voice control based on electrical equipment (TV, lighting, hot water, water heaters and other electrical equipment is controlled by the switch and the electrical equipment remote control). Uyghur phrase speech recognition process is composed of four parts (extraction of speech signal feature parameters for constructing acoustic model, build the language model and decoder part of the pattern matching. The technology related) of these four parts were compared and analyzed, with the choice of MFCC, HMM and N-gram and algorithm research in addition, this paper described in detail the MFCC feature extraction method, HMM acoustic model and its related algorithm and N-gram language model and smoothing method. The acoustic model and language model to realize the integration of mobile phone applications to achieve voice speech recognition control module of the intelligent Home Furnishing system in intelligent mobile phone system in Android finally. A voice recognition module through the intelligent Home Furnishing hardware connected with the mobile phone WiFi wireless communication protocol API to realize the control of electrical equipment. The application of Uygur speech control smart home.

【学位授予单位】：新疆大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TU855;TN912.34

【相似文献】