Self-recovery scheme for audio restoration using auditory masking

Area of Science:

• Digital Signal Processing
• Audio Engineering
• Information Security

Background:

• Self-recovery schemes embed compressed data for tamper detection and restoration.
• Audio applications require high transparency, making distortion masking critical.
• Existing methods may not adequately balance recovery efficacy with perceptual transparency.

Purpose of the Study:

• To propose a novel audio self-recovery scheme.
• To leverage auditory masking properties for improved embedding and restoration.
• To meet transparency thresholds for real-time audio applications.

Main Methods:

• Utilizing auditory masking properties to select optimal frequencies for embedding distortion.
• Mapping selected frequencies from the Fourier domain to the intDCT domain.
• Embedding and extracting reference bits within the intDCT domain for signal restoration.

Main Results:

• The proposed scheme achieves a transparency threshold of -2 ODG (Objective Difference Grade), suitable for audio applications.
• Demonstrated effective restoration capabilities for tampered audio signals.
• Validated the efficacy of using auditory masking for frequency selection and intDCT domain mapping.

Conclusions:

• The developed audio self-recovery scheme effectively restores tampered signals while maintaining high perceptual transparency.
• The integration of auditory masking properties and intDCT domain mapping offers a promising approach for robust audio self-recovery.
• The scheme's performance meets the stringent requirements for online music distribution and speech transmission.