摘要
通过同享编码器的多相子带滤波／MDCT混合滤波器组作为MPEG Layer III心理声学模型的频谱分析，可以将模型计算量降低80％以上。然而由于MDCT变换的能量抖动现象的存在，简单的一阶预测会导致纯音度计算的严重偏差。采用二阶后向自适应格型预测器计算纯音度的方案，改善了这种简化算法的计算精度。
By sharing the Polyphase／MDCT hybrid filterbank as the spectrum analysis module of the MPEG Layer III psychoacoustic model， 80％ of the calculation complexity can be reduced． But due to the fluctuation of MDCT spectrum， simple predictor will lead to serious error in the calculation of tonality index． In this paper， an enhanced tonality index calculation scheme is proposed． By using a second order backward adaptive lattice structure predictor， the precision of the simplified psychoacoustic model is improved obviously．
出处
《电声技术》
北大核心
2001年第7期3-5,4,共4页
Audio Engineering
