Different weighted values are added based on the contribution of each class to the whole spectral envelope, and a weighted linearly transformation based on the GMM posterior probability is presented.

• 为了克服一般分类线性转换算法中分类不准确所带来的误差，本文引入了分类线性加权转换的策略，根据不同子类的转换函数对谱特性的贡献，赋予不同的加权系数，给出了一种基于GMM后验概率加权的线性转换算法。