免手持通訊系統的應用有免手持聽筒電話、視訊會議、遠距教學及遠端醫療等系統,此種通訊系統必需解決聲學迴音之問題。可適性濾波器提供這個問題的有效解決方法。 首先本篇論文介紹適應演算法LMS家族與聲學迴音消除架構,進而以德州儀器的TMS320C32的浮點DSP處理器,實現適應演算法的NLMS與MDF應用在聲學迴音消除上。最後我們將這些演算法在實作方面作實際的運算量分析、比較。我們發現NLMS演算法架構簡單而且容易實現,適合短迴音路徑。MDF演算法其架構較為複雜,但是對於長迴音路徑,有其運算量少的優勢。
Hands-free communications system finds applications in hands-free phone, video conferencing system, and distance learning/medicine system. Adaptive filtering algorithm provides an effective means for solving the acoustic echo problem associated with such system. In this thesis, we briefly review the algorithms of LMS family. The acoustic echo cancellation configuration is then presented. We implement the normalized LMS and the MDF algorithm with Texas Instruments TMS320C32 floating DSP processor. Finally, we study and analyze the number of operations required in implementation. Our study shows that the NLMS algorithm is simple in structure, easy to implement, and suitable for short echo path applications. The MDF algorithm is quite complicated, however, it is very fast in the long echo path applications.