Gammachirp滤波器组在语音特征提取中的应用
首发时间:2011-11-09
摘要:介绍了一种基于Gammachirp滤波器组的语音特征提取方法。通过Gammatone滤波器级联一个非对称补偿IIR滤波器实现Gammachirp滤波器,相对于FIR实现法计算量大大降低。将Gammachirp滤波器组用于过零峰值幅度(Zero Crossing Peak Amplitude,ZCPA)特征提取,并与RBF后端识别网络结合进行仿真实验。在不依赖声压强度的情况下,得到不同啁啾因子下ZCPA特征的识别结果。实验结果表明,啁啾因子 时较其他啁啾因子识别结果更好。
关键词: Gammachirp滤波器 特征提取 过零峰值幅度(ZCPA) 语音识别
For information in English, please click here
Gammachirp filter banks applied in speech feature extraction
Abstract:This paper introduced a method of speech feature extraction based on Gammachirp filter banks. The Gammachirp filter is shown to be implemented through a combination of a Gammatone filter with an IIR asymmetric filter, which largely reduces the computational cost compared with the FIR method. The Gammachirp filter bank is used to extract Zero Crossing Peak Amplitude (ZCPA) feature, which works as the input of the radial basis function (RBF) network. The experiments were carried on in the case of level-independence, and the results of different chirp factors were achieved. The experiment shows that the results at is better than other chirp factors.
Keywords: Gammachirp filter feature extraction Zero Crossing Peak Amplitude (ZCPA) speech recognition
论文图表:
引用
No.****
同行评议
共计0人参与
勘误表
Gammachirp滤波器组在语音特征提取中的应用
评论
全部评论0/1000