机构地区: 华南理工大学电子与信息学院
出 处: 《模式识别与人工智能》 2005年第4期469-473,共5页
摘 要: 提出了一种基于WDFT的PLP特征提取算法。该算法采用WDFT来提高语音信号低频部分的频谱分辨率,与传统的提取方法相比更符合人类听觉系统的特性。对0~9十个英文数字和E-SET集的识别实验结果表明,本文提出的特征提取算法比传统PLP算法具有更好的识别效果。 In this paper we propose a new Perceptual Linear Predictive (PLP) extraction algorithm based on the Wrapped Discrete Fourier Transform (WDFT). The new algorithm WDFT is used in the new algorithm to improve the low frequency resolution of the input speech, which is more consistent with the characteristics of human auditory system than the traditional method. Experimental results on 0-9 digits and E-SET show that the proposed algorithm can improve the recognition rates efficiently.