MPEG-4 AAC感知噪声替代算法的研究与改进

The Research and Improvement of Perceptual Noise Substitution Algorithm in MPEG-4 AAC

针对先进音频编码(AAC)中感知噪声替代(PNS)标准算法在对噪声进行检测时容易造成误判而导致编码音质恶化的问题,提出了一种新的噪声检测算法.该算法利用了噪声频谱比音调频谱更为平坦的特性,设置了帧和比例因子带的双重门限,以对噪声进行精确检测.另外,在PNS模块中加入了预编码模块以进一步提高PNS的编码效率.实验结果表明,与标准算法相比,该算法大大提高了噪声的检测精度,在不增加编码复杂度的同时获得了更高的编码效率.

When using ISO standard of perceptual noise substitution（PNS） algorithm in advanced audio coding（AAC）,it seems to cause misjudgment when detecting the noise,and results in quality deterioration of coded audio.A new noise detection algorithm is proposed to solve the problem.It employs the characteristic of noise spectrum that is more flat than tone spectrum,and sets the dual threshold for the frames and scaling factor bands to detect the noise accurately.In addition,pre-coding module is added into the PNS module to further improve the coding efficiency of PNS.Experiments show that,compared with standard algorithm,the proposed greatly did improve the accuracy of noise detection,and obtain higher coding efficiency without any increase in coding complexity.