適應性回音消除之研究

Abstract

由於房間殘響的關係,適應性回音消除器的係數通常高達數千個,因此將產 生收歛速度慢和運算複雜度高等問題.本篇論文利用可變收斂步伐( stepsize)的最小均方差法(LMS)來增快收斂速度,和頻率點適應濾波器( FBAF)來降低運算量與傳輸延遲.可變收斂步伐的最小均方差法對每個濾波 器係數施與不同的收斂步伐,以增快收斂速度. 頻率點適應濾波器則是將 最小均方差法轉換到頻域操作,以降低運算複雜度.由電腦模擬得知,在不 影響回音消除品質的情況下,上述兩種方法可達到增快收斂速度或降低運 算複雜度的目的. Due to the long reverberation time of a room, the order of an adaptive acoustic echo canceller can be up to several thousands. This results in slow convergence speed and high computational compexity. In this thesis, a variable stepsize LMS algorithm, with each tap-weight being adjusted using different stepsizes, is used to increase the convergence speed. The FBAF algorithm, a frequency domain LMS method, is investigated to reduce computational complexity and transmission delay. The results of computer simulations show that the above two methods, compared to others, work quite well in terms of convergernce and computational complexity without degrading the quality of the acoustic echo cancellation .