论文部分内容阅读
数学形态滤波是一种关于信号形状处理的非线性变换，它能简化信号、消除较小分量而保留信号的基本形状特征．本文基于数学形态滤波方法提出了两个分别在时域和频域提取语音信号基音周期的方案，在频域提取基音周期的同时还能提取出语音信号的谱包络。它们具有简单、直观和计算效率高等特点。由于数学形态滤波运算是并行的、局部的，新方案适于并行化处理和易于硬件化实现。实验结果表明，选择合理的数学形态滤波参数以及线性预测编码参数，能获得准确的语音信号基音特征。
Mathematical morphology filtering is a non-linear transformation of signal shape processing that simplifies the signal and eliminates the small components while preserving the basic shape characteristics of the signal. Based on the mathematical morphology filtering method, two schemes are proposed to extract the pitch period of the speech signal in the time domain and the frequency domain respectively. The spectral envelope of the speech signal can be extracted while extracting the pitch period in the frequency domain. They are simple, intuitive and computationally efficient. Because the mathematical morphology filtering operation is parallel, partial, the new scheme is suitable for parallel processing and easy hardware implementation. The experimental results show that the accurate phonetic features of speech signal can be obtained by choosing proper mathematical morphology filter parameters and linear predictive coding parameters.