Kuan-Lin Chu

SSTMark achieves unprecedented robustness in speech watermarking by embedding watermarks at the semantic level, outperforming traditional methods even under severe distortions.