Adaptive Filter by Using Segment Proportionate Extended Correlation LMS Algorithm in the Double-Talk Condition

Echo path estimation in echo canceling for teleconference system is a problem in double-talk condition. The correlation-processing algorithms were defined by the authors to solve this problem. In this paper, we proposed a new promotional algorithm with fast convergence speed - proportionate extended correlation LMS algorithm (PECLMS). The idea of the PECLMS algorithm is introducing an improved proportionate adaptation into the ECLMS algorithm. Furthermore, we improved the PECLMS algorithm with segment proportionate adaptation. The segment proportionate extended correlation LMS algorithm (SPECLMS) have faster and more stable convergence speed than the PECLMS algorithm. The computer simulation results support the theoretical findings and verify the robustness of the proposed SPECLMS algorithm in the double-talk situation.