面向实时通信的低延迟高质量音频编码算法

High Efficient Perceptual and Low Delay Audio Coding Algorithm for Real-time Communications

提出了一种新的低延迟高质量音频编码算法,主要面向交互式的实时通信。为了降低编解码算法延迟,采用了相对较短、长度固定的变换窗,从而大幅度地降低了算法延迟。同时,为了在高压缩比下获得高质量的音频,运用修正的离散余弦变换(MDCT)的分析方法,不仅降低了算法复杂度,同时也提高了分析的精确度。算法的提出弥补了传统音频编码算法和语音编码算法的不足,它不仅具有较低的编解码算法延迟,而且在编码效率、音质和算法复杂度等方面,可以与时下的其它高级感知音频编码算法相媲美。

A novel high efficient perceptual and low delay audio coding algorithm for interactive real-time com- munications is proposed. Instead of using traditional audio coding technology, a short analysis windows of fixed length for low-delay purpose is adopted. In the same time, for the sake of obtaining high compression ratios and high quality, the psyehoacoustic model and stereo coding are modified and improved to increase the coding efficien- cy, by combining MDCT into analysis to lower the algorithm complexity and increase the analysis precision. The proposed audio coding algorithm makes up the disadvantages in both traditional audio and speech coding algo- rithms, which has not only low algorithm delay, but also equivalent coding efficiency and audio quality when com- pared with other state-of-the-art perceptual audio coding algorithms.