A Resilient Hybrid Audio Watermarking Framework Against Signal Processing Attacks

This paper offers a resilient hybrid audio water-marking scheme that combines Spread Spectrum (SS) embedding with Discrete Wavelet Transform (DWT), Discrete Sine Transform (DST), and Singular Value Decomposition (SVD). Multilevel transformation precedes the insertion of the water-mark into perceptually important subbands of the host audio, using the strength of spread spectrum modulation and singular value stability. System parameters were optimized using bit error rate (BER) and signal-to-noise ratio (SNR) under 64 kbps MP3 compression. Results were tested for several different attacks, including low-pass filtering, resampling, quantization, addition of noise, and MP3 compression at several different bitrates. Experimental results indicate that the system provides a payload of 172.26 bps with an acceptable audio quality of 27.2 dB SNR. Comparative evaluations demonstrate that the proposed system provides a desirable tradeoff of robustness, imperceptibility, and payload capacity and operates competitively compared to other methods for most tested conditions. This overall performance, combined with an efficient computational cost, makes the scheme a pragmatic solution for real-world, offline copyright protection applications.