With the development of information society, audio steganalysis has attracted more attentions recently. Echo hiding method is one of the most challenging research fields. However, most of them can't solve the contradiction of capacity, robustness and imperceptibility, which restrict the development of the system seriously. Some of them have low capacity, and others have relatively high capacity at the cost of the robustness and imperceptibility. In this paper, we proposed a novel echo kernel based on Delay Matrix, increasing the capacity to 16 times. Because there is more than one echo in the actual voice, multiple echo kernels guarantee the better imperceptibility. We also embed parity check information without decrease the capacity, improving the robustness of the system. The effectiveness of our system is illustrated by theoretical analysis, simulation examples, and listening test. Though the coverage of our current system is still small, the results have shown that it is better than the other echo hiding system.
展开▼