A multi-channel stereophonic system in which different audio signals are produced and caused to converge upon a listening site in an enclosure at selectively spaced time intervals so as to enhance the realism of the sound impressed upon the listener. A first pair of auxiliary speakers are arranged in front of and above a pair of conventional main speakers with one of the auxiliary speakers facing the ceiling of the enclosure. Signals are transmitted from an amplifier in selected phase relationship to each of the auxiliary speakers, the sound produced at least by each of the first pair of auxiliary speakers being selectively reflected off of sound-reflective surfaces in order to physically separate the added sound produced by the auxiliary speakers from the conventional stereo speakers. One of the speakers of the first pair of auxiliary speakers is connected out-of-phase to the amplifier while the other speaker is connected in phase. A second pair of auxiliary speakers are arranged behind the ideal listening site facing a rear wall of the enclosure. One of the speakers of the second auxiliary pair is connected out-of-phase to the amplifier while the other speaker of the second auxiliary pair is connected in phase.