基于新型TNS启用机制的高效音频编码方法

Highly Efficient Audio Encoding Method Based on Modified TNS Enabling Algorithm

时域噪声整型能有效地抑制由量化带来的时域上的前回音(Pre-echo)现象。提出了一种结合时域与频域,利用Levinson-Durbin及检测前回音的双重标准作为TNS的启用机制。提出了一种基于TNS启用机制的编码策略来替代计算复杂的心理模型中动态窗的算法,并验证了其在存储和计算上的开销。

TNS （Temporal Noise Shaping）is a technique to more effectively mask the spread quantization noise temporally. In this paper, we proposed an efficient time-frequency threshold using both Levinson-Durbin algorithm and pre-echo checking to determine whether to enable TNS. Due to the same nature of window-switching technique, the computational burdensome psychoacoustic analysis is omitted, thus saving memory and calculation costs in codec implementations.