AVS-P10立体声编码算法的优化设计与实现

Optimization design and implementation of AVS-P10 stereo coding algorithm

AVS-P10是我国首部具有完全自主知识产权的移动音频编解码标准,但是其立体声编码存在重建声像不够稳定、编码复杂度较高的问题。根据参数立体声编码原理,采用声道下混和立体声参数提取及合成技术,以AVS-P10的核心编码器为基础,设计并实现了一种高效的立体声编解码方案。实验结果表明,同等码率下,优化后的算法比AVS-P10立体声编码算法的主观音质提升约10MUSHRA得分,编码复杂度下降幅度达到40%~60%,解码复杂度略有下降。

AVS-P10 is the first national standard for mobile audio codec with completely independent intellectual property rights, but its stereo coding algorithm has the problem of unstable reconstructed sound image and high computation complexity. A new stereo coding scheme is proposed by using the channel down-mixing, stereo parameter extraction and synthetic technology. The frame and key modules are designed and implemented. Experimental results show that both of the subjective quality and computing efficiency of the optimized algorithm are better than those of the original stereo coding algorithm of AVS-P10. The encoding complexity decreases to 40%~60%, and the decoding complexity decreases slightly.