CELP宽带语音编码中感知加权滤波器的新算法

心理声学模型已经成功地应用于音频编码技术中，将心理声学模型应用到CELP语音编码技术中，构造新的感知加权滤波器，可以有效地提高合成语音的质量，但如何减少其算法的复杂度是CELP算法的一个关键．提出了LSP系数插值和滤波器系数插值两种新方法，以替代子帧中的心理声学模型掩蔽曲线的计算和转化，有效地减少其算法的复杂度．实验结果表明：在语音质量基本一致的情况下，LSP系数插值和滤波器系数插值两种算法的平均编码时间与原来相比分别减少18％和53％．该算法具有一般性，可以推广到其他CELP编码器的心理声学模型应用中．

New algorithm of perceptual weighting filter used in CELP wideband speech codec.

The psychoacoustic model has been successfully applied in wideband audio coding.By incorporating the psychoacoustic model into CELP speech codec and constructing a new perceptual weighting filter,the quality of synthesized speech can be effectively improved.The fundamental obstacle to the application of the psychoacoustic model lays its complexity.This paper tried to solve the problem by designing new schemes for the application of the psychoacoustic model.Instead of computing and converting the masking cur...