Design Of Smart Home Interaction System Based On WEB Client And Speech Recognition

Smart home provides people with a smart,safe and convenient home environment and is growing rapidly.At present,It is mainly based on the combination of the upper computer application(APP)and the control system,which has the disadvantages of single humancomputer interaction mode,complicated control and poor hardware development performance.In view of the above shortcomings,this paper designs an improved smart home interaction system based on Web development and speech recognition technology.Firstly,from the design perspective of the system-on-chip,based on the Cortex-M3 soft core,a smart home control system is built on the Field Programmable Gate Array(FPGA)platform.The main control can realize the customized design of the peripheral module according to the actual functional requirements,and improve the flexibility of the smart home control system and the coordination ability of the software and hardware.Then,combined with Web development technology,a smart home Web client that can realize real-time multi-site interaction is designed.Compared with the traditional App control mode,it can be used in multiple terminals across platforms with free installation,and the maintenance is simple.Finally,combined with speech recognition technology,the smart home speech recognition interactive system is designed,which expands it's human-computer interaction mode.In the field of speech endpoint detection,this paper designs an adaptive short-term energy threshold endpoint detection algorithm with error correction mechanism based on the time domain characteristics of speech feature transformation.Compared with the traditional fixed-threshold endpoint detection algorithm based on short-term energy characteristics,the accuracy and noise immunity of speech endpoint detection are improved when the speech endpoint detection recognition speed is not much different.Compared with the interaction scheme of the LD3320 speech recognition chip,the operation process of speech recognition is simplified,and the correct rate of speech interaction is effectively improved by 7%.