DCT based Audio Steganography in Voiced and Un-voiced Frames

In this paper, a robust audio steganography method is proposed based on voiced and un-voiced frames. The key idea is to change the magnitude of Discrete Cosine Transform (DCT) coefficients of voiced and un-voiced frames separately. Taking advantage of voiced and un-voiced characteristics it is possible to embed more number of bits in unvoiced frames. The proposed method proves the high imperceptibility of 43.9dB measured in terms of signal to noise ratio, for a payload of 1.08Kbps. This paper provides the tradeoff between SNR and Embedding Bit Rate which facilitate to decide the imperceptibility requirement. The experimental results proves that the method has high capacity of 240 bps to 1800 bps and provides robustness against the common signal processing attacks such as re-quantization, re-sampling and additive white Gaussian noise