A Novel Bit-Plane Shifting Algorithm for Scalable Audio Coding

Bit-plane shifting is one of the strategies for providing Fine Grain Scalability (FGS) in scalable audio coding. It is still desirable to cater the network bandwidth, device capabilities and user preferences. In this paper, a Noise to Mask Ratio (NMR) based bit-plane shifting algorithm is proposed. It is different from just sequential bit-plane coding used in MPEG 4 scalable to lossless (SLS) audio coding, and also different from other existed strategies at lower bit-rate audio coding. It firstly transforms normal bit-planes corresponding with scalefactor bands to the priority bit-planes by a bit-plane shifting according to their NMR. And then, a simple sequential bit-plane coding is used. The priority bit-plane means that a bit-plane corresponding to a scalefactor band which have larger NMR takes a higher priority of coding. Both the objective and the subjective audio quality evaluations give us good scores by comparing with the sequential bit-plane coding of SLS and Just Noticeable Distortion (JND) method. Otherwise, no extra side information is needed and the computational complexity is reduced substantially. The proposed NMR-based bit-plane shifting algorithm obtains good performance in both subjective and objective assessments while leaving the computational complexity at a low level.