Research And Implementation Of Acoustic Feedback Suppression Algorithms Based On Adaptive Filtering

The main function of the sound reinforcement system is to amplify the sound signal,and the characteristics of a sound field environment in the sound reinforcement system in which the microphone and the speaker are located together form a closed signal loop in the system,thereby generating an acoustic feedback phenomenon.The early methods of acoustic feedback suppression mainly include frequency and phase modulation,gain reduction and spatial filtering and they are not very ideal for suppressing acoustic feedback.With the development of hardware technology,an adaptive feedback-based acoustic feedback suppression method has been proposed,which can be applied to the initial stage of acoustic feedback signal generation and has good performance.The research topic of this thesis is an acoustic feedback suppression algorithm based on adaptive filtering.An improved acoustic feedback suppression algorithm based on adaptive filtering is proposed and the effectiveness of the algorithm is verified by simulation and experiment.Firstly,the mathematical model of the sound reinforcement system is constructed,including acoustic feedback path modeling and electroacoustic forward path modeling,besides,the characteristics of acoustic feedback establishment process and acoustic feedback signal are analyzed through experiments.Then,two adaptive filter algorithms based on mean square error are introduced: LMS(Least Mean Square)algorithm and NLMS(Normalized Least LMS)algorithm,and a new variable step size adaptive filtering algorithm based on Sigmoid function is proposed,which is called SVS-VMLMS(Sigmoid Variable Step Variable Momentum Least Mean Square)algorithm.This algorithm can make the filter achieve faster convergence speed at the beginning,and improve the stability performance of the system in the later stage.Then two de-correlation algorithms are introduced to reduce the correlation between the sound source signal and the system output signal as well as two kinds of indicators used to quantitatively evaluate the acoustic feedback suppression algorithm.After that,the algorithm simulation system is built base on MATLAB platform,and three kinds of acoustic feedback suppression algorithms are simulated.The simulation results show that the acoustic feedback suppression algorithm based on LMS algorithm,NLMS algorithm and SVS-VMLMS algorithm achieves the improvement of system gain of 8.978 dB,9.218 dB and 9.374 dB respectively,and the performance based on SVS-VMLMS algorithm is better than LMS algorithm and NLMS algorithm;Finally,based on the above research,the signal processing board based on TMS320C6748 DSP chip is designed and fabricated.Then based on the signal processing board,a experimental platform of sound reinforcement system is built.After that three adaptive acoustic feedback suppression algorithms are implemented by software,and measured the performance of them.The experimental results show that the acoustic feedback suppression algorithm based on LMS algorithm,NLMS algorithm and SVSVMLMS algorithm achieves 6dB,7.9dB,and 8.7dB system gain improvement respectively.The performance of acoustic feedback suppression algorithm based on SVSVMLMS algorithm is better than LMS algorithm and NLMS algorithm.The experimental results are consistent with the theoretical analysis of the algorithm and the results of the algorithm simulation.