QIM-Based Audio Watermarking with Combined Techniques of SWT-DST-QR-CPT Using SS-Based Synchronization

Abstract

Audio watermarking is a technique of inserting information into an audio file so humans are not aware of the existence of such additional information. In this paper, we propose synchronization of audio watermarking with Quantization Index Modulation (QIM) with combined techniques of Stationary Wavelet Transform (SWT), Discrete Sine Transform (DST), QR Decomposition, and Cartesian Polar Transform (CPT). Firstly, after reading host audio, we insert header using SS-based synchronization. Next, host audio is decomposed by SWT, then several subbands with average power above than threshold are selected for next process. Next, the selected subbands in time domain are transformed by DST into frequency domain. Then, the frequency domain signal is decomposed by QR decomposition. In R matrix, its coefficients at position (1,1) and (2.2) are transformed using CPT. Finally, QIM inserts watermark bits into CPT coefficients. Watermarking scheme has an average SNR value of 32.718 dB and 82.61% success rate in Stirmark Benchmarck for Audio (SMBA) as standard attack for audio watermarking with $\mathbf{BER} < 10\%$.