The thesis designed and implemented a complete solution for smart speaker system based on DuerOS,this system has the function of voice interaction and controlling the home device.The thesis also designed and built a MEMS microphone with a special structure of the vibrating membrane.First,the thesis disassembled and analyzed several mainstream smart speaker products.After that,the thesis designed and developed the solution from both hardware and software,and explained the design ideas and implementation methods in detail.The hardware solution mainly includes the following parts: MEMS microphone array,Raspberry Pi,ZigBee module.This paper analyzes each part in detail from hardware structure and implementation method.The thesis analyzed each part from the hardware structure and implementation methods.The software solution mainly includes the following parts: voice wake-up,DuerOS client for Linux,NLP and remote control based on ZigBee.And I analyzed each part from the design architecture for coding.The thesis also designed a Poly-based condenser MEMS microphone with two special structures,and described the preparation process of this MEMS microphone.Then I tested the acoustic performance of samples.The results show that its sensitivity and SNR(signal to noise ratio)are in line with industry standards,and it has high reliability in high temperature,low temperature and high humidity environment.Therefore,it was verified that this MEMS microphone with special structures has a good application prospect in the smart speaker system.Finally,the sample machine was built according to the software and hardware solution designed in this paper,and the sample machine was tested for functions such as voice wake-up,preset content service,custom developed voice application,and voice control application,the test results are in line with expectations.. |