A method for detection and localization of multiple signals in spatially colored noise by an arbitrary passive sensor array is presented. The method also enables exploitation of prior knowledge that the signals are uncorrelated so as to improve the performance and to allow detection and localization even if the number of signals exceeds the number of sensors. The estimation, based on the generalized least squares criterion, is both consistent and asymptotically efficient. The detection is performed via the minimum description length (MDL) principle and is proved to be consistent. Simulation results confirming the theoretical results are included.