Perceptual Audio Hashing Using RT and DCT in Wavelet Domain

Perceptual audio hashing, which summarizes big audio data into compact and robust digest, provides a useful tool for the identification, retrieval, authentication, and other information processing of multimedia contents. This paper proposes a perceptual audio hashing algorithm based on Radon transform in wavelet domain. The wavelet approximate coefficients are mapped into a two dimensional matrix. Radon transform is applied to extract the efficient acoustic features from the matrix, and then discrete cosine transform was applied on several random Radon projections to yield low dimensional feature vector and conduct hash sequence. The experiments were conducted on two databases: a speech database and a music database. The results demonstrate the efficiency of the proposed scheme in terms of discrimination, perceptual robustness and identification rate.