This thesis deals with optimization of sound projectors, based on knowledge on the auditory perception. In sound projection it is desired that the lagging projected sound dominates the localization. One of the most limiting factors here is the leading direct sound, which, however, can only be reduced to a limited extent since the focusing capabilities of sound projectors are physically limited. In order to enable the perception-based optimization, it was therefore essential to gain an understanding of the perceptual role of the direct sound in achieving localization dominance of the projected sound, and which perception-based requirements for sound projection result from this role. A review of existing literature on the perception in scenarios with leading and lagging sound revealed that further insights into lag localization dominance were needed to this end. These insights were gained by conducting several psychoacoustic investigations in an anechoic chamber, reproducing the sounds via individual loudspeakers. Lag localization dominance seemed to be strongly influenced by the temporal characteristics of the playback signal. Afterwards, comprehensive perception-based requirements for sound projection were derived and their consequences for the design of sound projectors were discussed. On this basis, a method for the perception-based optimization was developed with the goal to reduce the influence of the direct sound on localization. This method was named localization masking. Localization masking is based on the additional generation of one or more sounds arriving earlier and from another direction than the direct sound at the position of the listener. An investigation under laboratory conditions, using cascaded lead-lag pairs representing the sounds involved, suggested that localization masking has the potential to achieve that goal. Localization masking enabled the initial lag, representing the projected sound, to dominate the localization up to a 7 dB higher level of the initial lead, representing the direct sound. Finally, localization masking was investigated under realistic conditions. Localization masking was applied to real sound projectors in a real room and proved to work. Localization masking enabled a given projector to be effectively used with a playback signal that requires stronger focusing capabilities. Furthermore, localization masking enabled a projector with less strong focusing capabilities to be effectively used with a given playback signal.