Pleasant audio watermarking based on musical arrangement techniques focused on preventing musical unpleasant

In audio watermarking, preventing quality degradation after embedding is a fundamental requirement. Existing methods typically minimize the energy of the difference signal between the host and stego signals. However, once perceived, this difference is inevitably recognized as unpleasant noise. To address this issue, this paper proposes a novel watermarking method that embeds watermark sequences by using chord changes, specifically diatonic chord arrangements. The proposed approach is not based on human performance but targets desktop music production (DTM), which refers to music production using a PC. The primary goal of this research is to ensure that, even when the difference signal is perceived, it is recognized as a pleasant, harmonized musical component rather than noise. Experimental evaluations using MIDI data demonstrated the effectiveness of the proposed approach. First, a bit error rate (BER) of 0% was achieved for all test data, confirming complete and accurate watermark extraction. Furthermore, subjective evaluation using Mean Opinion Score (MOS) based on musical criteria showed that the proposed method scored, on average, 1.19 points higher than the comparative method, demonstrating superior auditory quality. However, the embedding capacity for certain test tracks remained below 100 bits. Therefore, a key future direction for future work is to increase the embedding capacity, for example by incorporating chords beyond diatonic ones or optimizing note placement techniques, while preserving overall audio integrity and quality.