Although miniaturized acoustic localization sensors have significant advantages in many fields, such as video conferencing, smart homes, intelligent scene monitoring, and robot audition, it is challenging to achieve high localization accuracy using them. In this study, a miniaturized acoustic localization sensor called SuperSoundcompass is developed. The system consists of seven identical microelectromechanical systems microphones distributed on a circular substrate with a radius of 4.5 cm. A high localization accuracy can be obtained with a limited number of acoustic sensors with compressive sensing applied. In order to verify the effectiveness of SuperSoundcompass, a series of simulations and experiments are performed under various conditions, such as low signal-to-noise ratio, coherent arrivals, and snapshot-starved data. The average root-mean-square error of SuperSoundcompass at 0 dB is 1.81∘. The results confirm its excellent performance in terms of accuracy.