摘要
本文介绍了传统的语音信号短时特征参数表征方法，并就语音信号的非平稳分析方法进行研究，重点分析了频率切变分布（ＦｒｅｑｕｅｎｃｙＳｈｅａｒＤｉｓｔｒｉｂｕｔｉｏｎ）、自适应旋转投影分解（ＡＯＰ）以及小波变换等方法，并且就各种特征表征的应用和性能进行了讨论。
オn this paper, the conventional methods of describing speech features are introduced and some methods of nonstationary analysis of speech signals are studied. In particular, the frequencyshear distribution, adaptive oriented orthogonal projective decomposition and wavelet transformation are discussed in detail. Additionally, the application and the performance of the methods are also discussed.
关键词
语音特征
线性预测编码
LPC
AOP分解
特征表征
speech features, linear predictive coding, timefrequency distribution,frequencyshear distribution, adaptive oriented orthogonal projective decomposition
