A Comparative Study of Lossless Audio Coding Schemes

Abstract

In this paper, we present a comparative study of different lossless audio coding schemes, which are implemented using different integer transforms. The audio signal under consideration, which is assumed to be integer-valued, as in the case of fixed-point implementation, is first decorrelated using the appropriate integer transform. The resulting integer coefficients are then entropy-coded to produce the output stream. Several integer transforms have been considered, such as the integer wavelet transform (IntWT) with different decomposition levels as well as different filters, the integer discrete cosine transform (IntDCT), and the integer Walsh Hadamard transform (IntWHT). Arithmetic and Huffman coding have been used for entropy coding. The performed simulation provides insight on the performance of the different integer transforms in the lossless audio coding context