New echo embedding technique for robust and imperceptible audio watermarking

Echo watermarking marks audio signals by adding an echo. The embedded signal is not additive noise but an audio signal having the same statistical and perceptual characteristics. The echo is even perceived as added resonance. Conventional echo watermarking techniques often exhibit inherent trade-offs between imperceptibility and robustness. A new echo embedding technique is proposed. The proposed method enables one to embed high energy echoes while the host audio quality is not deteriorated, so that it is robust to common signal processing modifications and resistant to tampering. This is possible due to echo kernels that are designed based on psychoacoustic analyses. Subjective and objective evaluations confirmed that the proposed method could improve the robustness without perceptible distortion