Research On Sound Data Clustering Method For Epidemic Early Warning

The research topic of this article is "Sound Data Clustering Method for Epidemic Early Warning".The main research purpose of this article is to investigate the trend of respiratory epidemic outbreak in a certain place through the collection and analysis of crowd-level cough data signals.The specific implementation scenario is to arrange a microphone in a crowded place(such as a laboratory,classroom),collect audio signals with coughs of different pe ople,and accurately count the number of people who cough in a segment of audio signals through the audio clustering method.This indicator is used to warn the trend of the outbreak of respiratory epidemics.The main research content of this paper is a system built based on research purposes.The system collects cough audio signals and is then divided into three modules.The first module is a data preprocessing module.Th e function of the module is to divide the collected audio signals into clusters to be c lustered.In each section,the second module is a data clustering module based on the ILP algorithm,and the third module is a data clustering module based on the twin n eural network.The audio features combined with the ILP algorithm and the twin neural n etwork are the two main types used in this article.research method.The cough audio data set is collected from a small program developed and Youtube website,and uses t he Noise-X92 noise library to simulate audio data with different signal-to-noise ratios to verify the anti-noise performance of the system.According to the clustering results obtained in the final experiment,when the signal-to-noise ratio is high,that is,when the surrounding environment is relatively quiet,the two types of methods can achieve high clustering accuracy.When the signal-to-noise ratio is low,the twin nerves The network method has more significant advantages.At the same time,the twin neural network also has its weaknesses,that is,it requires a large number of training data sets.The labeling of the data sets is labor-intensive and difficult,and the experiment time is longer.The research method of the ILP algorithm does not require a training data set,and the experiment time is faster.Therefore,the two types of metho ds have their own applications.