Overcomplete Dictionary Pair for Music Super-Resolution

With the development of society, people are going to pursue high quality music (HM). However, the Internet is filled with more generic music (GM) rather than HM. In this paper, we propose a novel algorithm called overcomplete dictionary pairs (ODP) algorithm. The ODP algorithm is able to recover HM from GM. The basic idea is to determine a set of elementary functions, called atoms, that efficiently capture music signal characteristics. There are mainly two steps. First, we employ K-SVD algorithm to jointly learn the overcomplete dictionary pairs of HM and GM frames. Then, we recovery HM via sharing the sparse representations of HM and GM. In order to validate the effectiveness of the ODP algorithm, the segment Signal-to-Noise Ratio performance and another objective parameter, perceptual objective listening quality assessment, which is introduced by the ITU are concerned. Experimental results show that the proposed algorithm is much better than conventional reconstruction method like shape-preserving piecewise cubic interpolation.