Research And Design Of The Howling Suppressor Based On FPGA

In the sound reinforcement system,howling is an acoustic phenomenon which is the most common and difficult to be completely eliminated.Howling comes from the self-oscillation of sound reinforcement system.Its conditions of establishment are very easy to meet and it is easier to occur when the pickup device is sensitive and the gain of the amplifier is strong.The impact caused by the howling is extremely bad.It will lead to distortion of the sound and destruction of the atmosphere at the scene if the impact is little and may damage the sound amplifying system and even cause harm to the human auditory system if the impact is serious.Therefore,it is of great significance to suppress the howling in the field of sound reinforcement system.How to effectively suppress the howling without reducing the sound quality has been plagued the majority of engineers and technicians because of the characteristics of easy to build and difficult to suppress.The traditional method of howling suppression is passive and has great limitations.The digital howling suppressor has gradually replaced the traditional method of howling suppression with the rapid development of the digital signal processing technology.Among them,the adaptive howling suppressor,which is characterized by the rapid suppression of the whistle,little damage to the sound quality,and the automatic adjustment of the suppression parameters,has become the mainstream.The most outstanding howling suppressors in the market are produced from abroad at present.They are expensive and their popularity is low.So it is an urgent task to study the cost-effective howling suppressors of our own.This thesis makes a deep analysis of the characteristics of the howling by the actual acquisition of the howling audio.Then,the wavelet packet transform algorithm is chosen as the howling suppression algorithm according to the characteristics of the howling.Wavelet packet transform is an algorithm based on wavelet transform,whose decomposition method is more precise and more flexible than wavelet transform's.The howling suppression scheme adopted in this thesis is decomposing the audio signal by multilayer wavelet packet firstly,finding out the howling node according to the energy of wavelet packet node then,reconstructing the node by wavelet packet and inversing the result of reconstruction and adding the last result to the original audio signal to suppress howling finally.The feasibility and accuracy of the scheme are verified by simulation in MATLAB.This thesis chooses to implement the scheme of howling suppression on FPGA and designs a high-performance howling suppressor because of the high speed and parallel processing ability of FPGA.The design is divided into two parts: hardware design and software design.The part of hardware designs the hardware platform of howling suppressor.The part of software adopts the top-down design idea to design and simulate each module separately,and finally completes the design of the whole howling suppression system.The software system is downloaded to the hardware platform for verification,and the design of the howling suppressor based on FPGA is completed at last.