Computationally efficient algorithm for narrowband active sound profiling

In active sound profiling (ASP), the target is to obtain a certain sound field or profile using similar techniques as in active noise control (ANC). Active sound profiling can be applied to a passenger car, for example, to modify the engine sound in the cabin. A fundamental algorithm in active sound profiling is the command-FXLMS (C-FXLMS) algorithm, which is an extension of the famous FXLMS algorithm widely used in active noise control. The computational demand of the C-FXLMS algorithm is dominated by the reference signal filtering. In order to reduce the computational burden, an alternative method to modify periodic reference signals has been developed. Since the reference signals are periodic, a method delaying the signals and modifying their magnitude can be used. The magnitude and phase delay values are obtained via the system identification process and stored in lookup tables. The new method is computationally efficient but requires more memory than the conventional filtering. In this paper, the C-FXLMS algorithm has also been combined with the eigenvalue-equalized FXLMS (EE-FXLMS) algorithm. In the EE-FXLMS algorithm, the secondary-path model is modified so that the magnitude becomes flat. This leads to frequency-independent step sizes in the adaptation of the algorithm, increasing robustness and enabling easier tuning of the step sizes. The memory requirement is also decreased since the size of the lookup table for magnitude values is significantly reduced. The eigenvalue-equalized C-FXLMS algorithm using the lookup-table based reference signal compensation has been tested in a simulation model. The model is based on an experimental ASP system installed in a passenger car. Simulations have been carried out both in ASP and ANC. Results prove that the algorithm converges fast and, in ASP, is able to track the desired levels with sufficient accuracy. In ANC, the engine sound is significantly attenuate