一种基于多描述编码技术的高质量音频编码算法

A New High Quality Audio Coding Algorithm Based on Multiple Descriptions Coding

本文依据感知音频编解码基本原理,研究和设计了一种基于多描述编码技术的高质量音频编码算法。这种算法具有较好抗丢包性能,算法的总体思路是先在分析与合成的层面上把音频分解为听觉掩蔽门限和剩余信号,然后在量化和编码层面上分别对音频的听觉掩蔽门限和剩余信号进行多描述处理。结果表明,在所提出的多描述抗丢包音频编解码算法框架下,多描述算法的抗丢包性能明显优于单描述的抗丢包性能,标量量化多描述算法的抗丢包性能比奇偶分离双描述算法和对偶变换双描述算法的抗丢包性能都要好。

According to the general principles of perceptual audio coding,a new audio coding algorithm based on MDC is studied and designed. This algorithm has good anti-packet-loss performances,the main ideas of it are the decomposition of audio signals into hearing masking thresh old and residual signals in term of audio analysis and synthesis,followed by MDC processing of both hearing masking threshold and residual sig nals in term of coding and quantization.The results have shown that,in the frame work of MDC anti-packet-loss of audio coding algorithms p ropesed in this paper,the anti-packet-loss performance of audio multiple descriptions algorithm is obviously better than that of single descripti on algorithm,two descriptions scalar quantizing algorithm is better than two descriptions pairwise correlating transforms algorithm and two descr iptions odd-even separating algorithm, and two descriptions scalar quantizing algorithm is same as three descriptions scalar quantizing algorithm.