Audio Steganography Using Bit Modification - A Tradeoff on Perceptibility and Data Robustness for Large Payload Audio Embedding

Audio steganography using bit modification of time domain audio samples is a simple technique for multimedia data embedding with potential for large payload. Depending on the index of the bit used to modify the samples in accordance with the data to be hidden, the resulting stego audio signal may become perceptible and/or susceptible to incorrect retrieval of the hidden data. This paper presents some results of the tradeoff between the conflicting requirements of data robustness, payload and imperceptibility. Experimental results on both clean and noisy host audio signals indicate that while the payload can be as high as over 3000 bits/s much higher rate than common audio data embedding techniques noticeability of embedding is decreased and noise tolerance increased by using higher bit indices than the traditional least significant bit. Bit error rates of below one percent were observed for data retrieved from noise-added stego audio signals with 39 dB of SNR for an embedded payload of over 10 Kbits in a 3.3 s host audio.