Jörgen Nordberg; Joakim Skoog MEE-96-01, pp. 13. Dept. of Signal Processing, 1996.
This Masters thesis report evaluates a method of acoustic echo cancellation in a hands-free mobile phone system. The method uses an adaptive subband filter architecture to get a faster convergence rate and adaptive FIR filters that are easy to handle. The filter bank that split the signal is employing a shorter FIR-filter than what i suggested in the literature. This has resulted in improves echo suppression and faster convergence rate.
Simulations in Matlab and implementation in an DSP (Digital Signal Processor) has been performed. An echo cancellation of about 32 dB was achieved from simulations, the used signals were real signals recorded in a car. The simulations show a faster convergence rate compared to a fullband LMS-flter.