Research On Recognition Of Sound Signals In Complex Background

The rapid development of society to bring a lot of data and information,and how to find useful information is particularly important,and for sound data,if it can effectively collect,process,identify the data,it becomes more valuable.In this paper,the study of complex background of the sound signal recognition system,and the system in-depth analysis and demonstration.The research on feature extraction method and recognition technology is focused on,and it is possible to realize the accurate sound signal recognition system under complicated background.In this paper,three types of sound signal research,including Ferrari car horn,Lamborghini car horn sound and horses horse horn sound.Through the data acquisition platform,using the GRAS microphone to collect the car horn sound,through NI-9234 data acquisition card and NI-DAQmx VI sub-module in LabVIEW,the sound signal to the image display,and save as.Wav format,and then transfer the data to the computer.In this paper,we focus on the extraction and identification of sound signals under complex background.The software of sound signal recognition is designed by using the MATLAB user GUI development program.The details are as follows:Firstly,endpoint detection and feature extraction are performed on independent sound signals.In the time domain for endpoint detection,according to the characteristics of the research object,a double domain feature extraction method is proposed to extract the feature of the signal.Secondly,the sound segment with the signal to be measured is pre-processed,that is,noise reduction and blind source separation.The 5th order Butterworth low-pass filter is used to denoise the signal to be measured,and then independent component analysis(ICA)algorithm to separate the sound signal.Then,an improved method for increasing the momentum term and the adaptive operator is proposed for the problem that the traditional BP neural network recognition algorithm in voice recognition is not ideal in the convergence rate and so on.Finally,from the practicality and integrity of both sides to develop and design a complex background ofthe sound signal recognition software system.For the complex background of the sound signal recognition system,in this paper,a feature extraction method,which is suitable for the object of this paper,is proposed by combining the characteristics of the object,namely in the time and frequency domain analysis at the same time,establish and improve the characteristic parameter values,improve the limitations of predecessors feature extraction method,the simulation results show that the proposed method can extract the eigenvalues of the signal more effectively and the effect is better.In this paper,the recognition algorithm is the improvement of the traditional algorithm optimization,by updating the network weight formula,and compared with the traditional algorithm,it is found that the convergence rate of the improved algorithm is reduced and the recognition rate of the system is improved.