An enclosure expansion type virtual engine sound system is provided. The system includes an enclosure case having a case enclosure accommodating a speaker to which a virtual engine sound is output and an enclosure back beam with an internal space divided into a speaker enclosure and a expanded enclosure as an extended space of the case enclosure for low band amplification of the engine sound and a space in which the engine sound is output in three directions. Accordingly, the enclosure back beam implements a low band amplification function in addition to a crash absorbing function. The engine sound is output in three directions through an internal space of the enclosure back beam, thereby solving a space restriction of a narrow engine room while satisfying regulations of the speaker.