Research And Implementation Of Howling Suppression Algorithm For Conference System

The howling is common phenomenon in the conference,the generation of howling will not only reduce the gain of sound reinforcement system,which cause can’t get the conference minimum pressure level,but it also can damage the sound reinforcement system equipment in the conference,what is worse,it is causing a certain hazard to the human ear.Aiming at the single function of the existing conference howling suppressor,entrusted by Xi’an Rui core company to commissioned the design of the howling suppression function module on the basis of the ADI platform,and with the performance index of howling suppressor on the mark,such as resolution,increase gain,the reaction time is compared to determine the performance of the design.After studying the existing howling suppression algorithms and mature products,this paper analyzes the process of howling and the characteristic of howling signals,and then the algorithm and implementation process of howling suppression is proposed.The subject of howling suppression algorithm mainly includes two blocks: compensation of frequency response characteristics of sound reinforcement system and suppression of howling based on notch filter.Aiming at the not flat of sound reinforcement system spectrum,and peak of the spectrum is more likely to produce howling.The algorithm of the subject design first compensates the frequency response of the sound reinforcement system spectrum.the multisampling rate FFT transform is calculated first,and then compute 31 sections of frequency spectrum.According to the spectrum characteristic,the parameter equalizer is set to compensate the frequency response characteristic automatically.Notch suppression howling system which includes the frequency detection and the notch filter design,the howling frequency detection is based on the research of the characteristics of the howling signal,and the design of the previous algorithms is improved.Reducing frequency error by primary frequency interpolation correction and DTFT quadratic correction.After finding the howling frequency,set up nine fixed and three dynamic ultra-narrow band notch filter to suppress howling.The gain of the filter is variable,in order to prevent excessive filter and excessive gain,which it will affect speech.Finally,on the ADI platform required by the company,combining with the laboratory equipment,this paper selects the SHARC-21489-EZ-KIT evaluation suit which has more application in audio processing.The algorithm is transplanted to DSP for software and hardware simulation,and the algorithm is improved and optimized by the test results.The test result is compared with the performance of the professional howling suppressor.In the same environment,the performance of the project howling suppression scheme better than the feedback suppressor.In the performance of the project,the resolution of the scheme is 1Hz,at least the sound reinforcement system gain of the conference room is increased by 8 dB,and the reaction time is within 0.3s,which achieves the expected effect.