Inaudible Speech Watermarking Based on Self-compensated Echo-hiding and Sparse Subspace Clustering | IEEE Conference Publication | IEEE Xplore