Advanced residual coding for MPEG surround encoder

Abstract

MPEG Surround (MPS) has been widely known as both efficient technique for encoding multi-channel audio signals and rich-features audio standard. However, the generation of residual signal in the basis of a single module in MPS encoder can be considered as not optimal to compensate for error due to down-mixing process. In this paper, an improved residual coding method is proposed in order to ensure the down-mixing error can be optimally minimised particularly for MPS operation at high bit-rates. The distortion introduced during MPS encoding and decoding processes is first studied which then motivates for developing an approach which is more accurate in determining residual signals for better compensation for the distortion. A subjective test demonstrates that the MPS with improved residual coding can be competitive to Advanced Audio Coding (AAC) multi-channel for encoding 5-channel audio signals at bit-rates of 256 and 320 kb/s.