Experimental results show that this method can obtain better results than the Gaussian mixture model (GMM) and Fuzzy vector quantization method in the case of the little training data.

• 实验表明，该方法对于较短的训练语音，其识别效果优于高斯混合模型和模糊矢量量化。