A DCT–SVD-Based Speech Steganography in Voiced Frames

Abstract

A novel speech steganography method is presented in this paper by cascading discrete cosine transform and singular value decomposition. The significance of the proposed method is that it utilizes the voiced part of the speech signal, making it highly imperceptible and robust. Imperceptibility of the proposed method is quantified by both signal-to-noise ratio and subjective listening tests. Our experimental results of bit error rate demonstrate that the proposed audio steganography method is robust against MP3 compression, AWGN, re-sampling and re-quantization attack with high data payload. The undetectability of proposed method is verified using a powerful steganalysis tool, and the results are compared with a well-known steganographic technique.