EUSIPCO'2002 banner

Paper data
Pitch analysis-based acoustic echo cancellation over a nonlinear channel

Lu Xiaojian, McGill University, Canada
Champagne Benoit, McGill University, Canada

Page numbers in the proceedings:
Volume I pp 159-162

Room Acoustics and Noise Control

Paper abstract
This paper proposes a new acoustic echo cancellation (AEC) system which combines a conventional AEC algorithm with the pitch analysis technique. Departing from the use of pitch prediction in speech coding, we use two signals, i.e. the estimated echo and the residual echo, to fulfill the pitch extraction in the AEC system. This system is suitable for the applications of digital networks where the nonlinearities of the echo path significantly degrade the performance of conventional AEC algorithms. Simulation results using a nonlinear channel show that, compared to conventional AEC algorithms, the proposed system remarkably suppresses the acoustic echo during double-talk.

A PDF version is available here