Research On Howling Suppression Algorithm In Sound Reinforcement System

Howling is a common phenomenon that occurs in sound reinforcement systems such as stadiums,classrooms,and large-scale performances.The sharp and harsh howling phenomenon not only limits the performance of the sound reinforcement system,but also makes the sound reinforcement effect of the sound reinforcement system far less than expected.Affecting the hearing effect,the severe howling sound may also burn out the amplifier and other units in the sound reinforcement system,causing certain damage to the human ear.Therefore,howling suppression is an important subject in sound reinforcement systems.The research and implementation of corresponding howling suppression algorithms have a very important impact on the development of sound reinforcement systems.This article aims at the above problems,and takes the existing howling suppression algorithm as the main research object.Based on an in-depth analysis of the existing algorithm principles,it designs an economical and practical howling suppression system to suppress the howling phenomenon and enhance the expansion of the sound reinforcement system.Purpose of sound gain and output sound quality.The main research contents of this article are:First,the principle of howling and the characteristics of howling are introduced.The basic principles of three howling suppression algorithms that are widely used in practical applications and several key steps in the implementation process based on the notch filter howling suppression method are studied.Several of the adaptive acoustic feedback cancellation methods are used to reduce decorrelation method of correlation between acoustic feedback signal and input signal.Then,based on the theoretical analysis of the existing adaptive variable step size algorithm,a new nonlinear function expression is constructed to replace the constant parameter in the G-SVSLMS algorithm,and a new improved variable step size is obtained.LMS algorithm.This algorithm can obtain smaller steady-state error when the convergence speed is fast,which improves the defect that the G-SVSLMS algorithm is in the steady-state phase,affected by noise,and the step size changes too quickly.At the same time,the algorithm is applied to the howling suppression system and simulation experiments.The results indicate that in the two environments of speech and noise,theimproved variable step size LMS algorithm compares with other variable step size algorithms,while ensuring the convergence speed.With smaller steady-state error,the algorithm has better performance.Lastly,built an ARM-based hardware platform.By comparing the advantages and disadvantages of the three algorithms,and considering the hardware cost required for the algorithm implementation,the final choice is simpler and more practical,and the effective method of the trap howling suppression method is implemented and tested on the hardware platform.Tested in the sound field environment of the laboratory,from the experimental results,we can see that the howling suppression system designed in this chapter can detect howling signals in a short period of time and suppress them simultaneously,which improves the maximum stable gain of the system and ensures the sound quality of the output audio signal with good results.