Design And Implementation Of Smart Speaker System Based On OpenWrt

In recent years,with the rapid development of the Internet of Things and artificial intelligence,people's daily life is quietly changing.People have begun to pursue a convenient,rich and smart home life.As a result,the field of smart home has become extremely popular.At the same time,for the increasing demand in the home environment,the user's investment cost for home equipment continues to rise,and the traditional home speaker equipment has limited audio resources,cannot communicate with existing home equipment,and cannot provide home users with personalized service.Home speakers show the characteristics of low versatility and insufficient personalization.Therefore,it becomes necessary to design a smart speaker system that saves costs,deeply integrates with the home scene,and has novel functions to meet this home demand.To this end,the work content of this paper is as follows:(1)After analyzing the overall requirements of the smart speaker,the overall architecture of the smart speaker system is proposed,then the hardware platform is designed,and the overall software system and speech emotion recognition solutions are proposed.(2)For the audio problem of smart speakers,the OpenWrt system environment is built and the WM8960 audio driver and audio player of smart speakers are implemented based on the embedded audio architecture to complete the audio system of smart speakers.(3)Analyze the user's requirements for software functions,design and implement the audio playback module,wireless communication and interaction module on the embedded end of the smart speaker device;develop the audio playback control APP on the mobile phone terminal.(4)For the user's physical and mental health monitoring function,the smart speaker speech emotion recognition system is designed.Including speech emotion feature extraction,genetic algorithm optimization support vector machine classifier and improved binary cuckoo search algorithm feature selection,etc.Finally,a speech emotion recognition system combining feature selection and SVM classification is designed and verified by experiments.(5)Based on the above design and implementation of the smart speaker system,the system was tested in sub-modules.After testing,the smart speaker system functions are basically intact and functioning normally,which can meet the needs of smart home scenarios and has certain market value and social significance.