Research And Implementation Of The Smart Home System Based On Specific Vocal Recognition

With the development of mobile Internet technology and the Internet of Things,smart homes are not only an important part of smart cities,but also will be the development trend of urban planning and construction of residential communities in the future.Now,although the smart home brings comfort and convenience to the user,the common smart home system needs to download the APP match devices,so that the smart home devices can be controlled and managed by manual clicking.To address the issue of complicated operation and security of smart homes,this thesis proposes a method that it can achieve recognition about person who speek now and content of speech to control the smart home devices based on the channel compensation algorithm and hidden Markov hybrid model(PLDA-HMM).The control of smart home devices enables users to get rid of their dependence on mobile phones and make the smart home more intelligent.The purpose of this project is to realize a smart home system for specific vocal recognition based on the wireless smart home environment.Firstly,for the specific vocal recognition,a PLDA-HMM hybrid model algorithm is proposed.The algorithm recombines the optimal parameter values of isolated words with the speaker's i-vector matrix,this matrix is a characteristic parameter that can represent the speaker and isolated words,then get the speaker and isolated worads corresponding to the maximum probability by iterative calculation.Secondly,Due to the transparency of interactive information about the smart home,only when the interactive information was acquired can the smart home devices be controlled.Therefore,an improved decision tree algorithm is proposed to obtain the smart home system control commands,from the acquisition of interactive information to the establishment of a decision tree model,realized successful acquisition of control commands for smart home devices.Finally,according to requirements analysis and design guidelines,two specific vocal recognition smart home design solutions were proposed and analyzed the advantages and disadvantages of the two design programs,completed the implementation and test verification of smart home systems for specific vocal recognition.JZ2440 development board as a hardware platform,building a Qt-based application software platform in Linux environment.Then connect the JZ2440 development board to the wireless router,the smart home devices were controlled by Real-time voice input.The test results show that the smart home devices were controlled by the smart home system that was designed in this thesis,So,it can meet the requirements.