Using Kautz filter for adaptive acoustic echo cancellation

A FIR filter is commonly used in an acoustic echo canceller (AEC) because of its simple structure. However, it cannot effectively cancel the acoustic echo due to its many required parameters. A time-varying room transfer function is physically due to its time-varying zeros and not its time-invariant acoustical poles. This physical reasoning motivates the proposal of using a Kautz filter as a possible solution to accurately describe an acoustic echo channel with few parameters. The low-complexity NLMS algorithm is used to recursively estimate the parameters. The adaptive Kautz filter converges to an echo return loss enhancement (ERLE) of 5 dB more than the adaptive FIR filter.