| Acoustic feedback is very common in public addressing systems, especially whenthe sensitivity of microphone is higher, and when PA gain of the system is larger. Theharm of howling is very great, there are three performs:firstly,interrupt the speech,which makes listeners uncomfortable; secondly, affect the gain of the PA system, inorder to avoid howling, the operator always can’t tune volume too high.thirdly,damaging the audio equipment, because the instant power of howling is far beyond thescope of the PA system can withstand.It is because the whistle is so common and it’s hazards are so serious, audioworkers devote much energy to solve this proplem, and get lots of results. Thetraditional method is to built a mathematical model to describe the sound field. but thesound field is changing,even when people moving in the house. So this way can’t workwell in fact. With the development of electronic technology, the scientists are trying tochange the sound characteristics to suppress the whistle, and engineering practice showsthat.Now there are some typical use, it seems like shifting frequency. All kinds of theseways can suppress howling to a certain extent, but they only be used in a certain field.As all above saying, our final goal is to develop a common method, which hassmall distortion, rapid response and low cost. Base on the successful application ofwavelet in mechanical failure detective and compression of digital voice, we proposethat using wavelet analysis to treat the howling signal, in order to pick up the howlingpoint and remove it. We propose arithmetic like this: firstly, we use wavelet divide thesound into some bags, then analyse the energy of these bags, pick up the most powerfulone which contains howling, then remove it. |