A t-out-of-n audio cryptography is a secret sharing scheme that can be used to hide a digital message into n sounds and human being's ears can perceive the message from any t of the n sounds playing simultaneously. In 1998, Desmedt et al. proposed the first audio cryptography scheme (ACS or AC scheme) by using the characteristics of the wave interference, i.e., in phase or out of phase. However, their scheme can generate 2-out-of-n ACS with log2(n) cover sounds only. In this paper, a new t-out-of-n ACS is proposed. The proposed scheme is to hide a secret message by a specified cadence of which each is played at either different or the same time in n music. The secret message can be perceived according to the number of the specified cadences by any t of the n music playing simultaneously. It can achieve any t-out-of-n scheme theoretically with only one cover sound. Furthermore, we show that there exists a transformation between a general VSS scheme and the proposed AC scheme. Any Boolean Matrices, which can be used to construct a VSS scheme, can also be transformed to construct the ACS, and vice versa.