| Sound field reproduction techniques can be used in order to generate the noise of a running car or plane inside a laboratory space.By means of this the sound quality of those vehicles can be evaluated in test rooms with a high level of real experience.This work aims at the comparison of the two sound field reproduction methods Least-Squares(LS)and Least-absolute shrinkage and selection operator(Lasso)based on simulation and experimentation.While the LS-based method is established in the field of sound reproduction,the Lasso demonstrates an efficient tool for reducing the loudspeaker number.For the simulation and experiments the reproduction is ill-posed and the reproduction error measured by different numbers of sampling points is analyzed.This work introduces the way of implementing both mentioned sound field reproduction methods and points out the determination of three different transfer function types,the Green function for free field condition,the Image method for nonfree field condition and the way of measuring transfer function.The simulation considering non-free field condition turns out,that the sound field reproduction of both methods is similarly accurate at a high number of sampling points.In the simulations and experiment reproducing the sound field at the locations of microphones,which means a low number of sampling point,the LS outperforms the Lasso for the same power level.As an application to reproduce the loudspeaker number combining both methods,the two-stage Lasso-LS algorithm is introduced.In the first-stage the Lasso selects a set of best positioned loudspeakers.This set is used for the sound field reproduction based on the LS method in the second-stage.The result shows,the lower the loudspeaker number,the lower is the accurateness of the sound field reproduction.However,comparing the results of the two-stage Lasso-LS algorithm with the singe-stage LS demonstrates,that a reduction of 6 loudspeakers for a total number of 20 loudspeakers is acceptable. |
PDF Full Text Request