Research And Application Of Speech Recognition Algorithm Based On Isolated Words

With the development of speech recognition technology,applying it to smart homes,controlling home appliances through voice signals,and realizing smart living have also become a very challenging problem.Aiming at this problem,this paper mainly researches the speech recognition algorithm based on isolated words,and applies it to the smart home control system to control the home equipment through voice commands.The main content of this article is as follows:(1)An improved endpoint detection algorithm is used.The voice signal is pre-processed,such as pre-emphasis,windowing and framing,etc.,and then endpoint detection is performed to reduce the amount of data.In order to improve the robustness of the algorithm and solve the problems of missed detection and false detection,this paper improves the traditional dual-threshold endpoint detection algorithm,uses an improved endpoint detection algorithm,and performs simulation analysis through MATLAB.The experimental results show that it is accurate The speech signal is detected,thereby improving the accuracy of speech recognition.After detecting the endpoints,the three methods are compared:linear prediction coefficient,linear cepstrum prediction coefficient,and Mel frequency cepstrum coefficient.Selecting the Mel frequency cepstrum coefficient to extract feature parameters of the speech signal has a good effect.(2)The layered matching algorithm is used to improve the traditional DTW algorithm.When the speech signal is recognized,a comparison experiment is performed between the traditional DTW algorithm and the HMM algorithm,which shows that the HMM algorithm takes longer to train and the algorithm is more complex,while the traditional DTW algorithm has lower accuracy and longer recognition time.DTW algorithm,and the use of hierarchical matching algorithm in pattern matching,by reducing the need to match the pattern,improve the rate and accuracy of the algorithm.This article selects commonly used household appliance command words for speech recognition.The experimental results show that the isolated word-based speech recognition algorithm used in this article is reliable,feasible,fast in recognition rate,high in accuracy,and can meet the requirements.(3)In this paper,the overall scheme of the voice smart home system is designed.The voice recognition algorithm is used in the smart home,and the home appliances are controlled by voice signals.The hardware design includes the choice of circuit design and communication mode;in software design,there are two modes of voice control and key control,with friendly interface and simple operation.