低码率语音编码中感知加权滤波器的一种新型算法

New Method of Designing Perceptual Weighting Filter Used in Speech Encoder

声学心理模型是对人听觉系统生理结构和人耳主观感知特性的数学抽象模型．这种模型已成功运用于宽带音频编码中．首次提出将声学心理模型原理应用于语音编码中感知加权滤波器的设计．通过对ＩＴＵＧ．７２３．１６．３／５．３ｋｂ／ｓ双速率编码算法中感知加权滤波器的改进，编码器的ＭＯＳ分可以改善０．１～０．３，而且新算法所要求的计算量仅比原算法大０．２６ＭＩＰＳ．基于声学心理模型的感知加权滤波器，与现有各类语音编码器中所用的感知加权滤波器相比，有自适应强、更符合人耳听觉特性、主观处理效果更佳的优点．

Psycho acoustic model is a mathematical model about human acoustic structures and human ear's physiologic characters, which has been applied in wide band audio coding successfully. The psycho acoustic model based perceptual weighting filter is developed. Such kind of filter is tested in G.723.1 6.3/5.3 kb/s dual rate speech encoder. The result shows the encoder's MOS score is improved by 0.1 to 0.3. Moreover, the new algorithm increases computing complexity only by 0.26 MIPS. Compared with the weighting filter now available, it is more adaptive to the input speech data, more coincident with human ear's physiologic character.