A Robust Design and Performance Analysis of Audio Data Hiding Technique on MPEG Surround

This study presents a robust design and performance evaluation of an Audio Data Hiding technique for MPEG Surround, a widely adopted standard for multichannel audio coding. Previous approaches typically embed secret messages into multichannel input signals before the MPS encoding stage, which often leads to synchronization issues and limited robustness against common signal-processing operations. This paper proposes a novel method that embeds messages directly into the MPS downmix signal, a mono signal generated during the encoding process. This strategy simplifies the embedding procedure and enhances compatibility within the MPS framework. The proposed method uses textual secondary data and is evaluated based on message extraction accuracy and perceptual audio quality. Comprehensive robustness testing constitutes the primary contribution of this work. The method is assessed under various distortion scenarios, including additive noise and audio compression. Experimental results demonstrate an average Bit Error Rate (BER) of 0.05 and a Signal-to-Noise Ratio (SNR) of 16.07 dB, indicating acceptable audio quality. Under more severe conditions, such as Gaussian noise and AAC compression, the method maintains a low average BER of 0.16%, confirming its resilience. These results verify that the proposed technique is both effective and robust for practical data hiding applications in MPS-based audio systems.