Key-Dependent Compressed Domain Audio Hashing

Perceptual audio hashing has been applied in audio content identification, retrieval and indexing for its accuracy and perceptual robustness. Security issue becomes important when perceptual hashing is applied in audio content authentication. A key-dependent compressed domain perceptual audio hashing algorithm for audio content authentication is proposed. A randomization scheme controlled by a random seed is applied in hash generation for random feature selection and post randomization. The hash function is key-dependent and collision free. Meanwhile, the randomization has little effect on robustness. The hash value is invariant to content preserving operations.