论文部分内容阅读
本文从最大后验概率估值的角度讨论了语音信号识别中的时间规整问题,讨论了由此导出的方法与常用的动态时间规整方法之间的关系,并对这一方法的普遍意义及其在语音识别中的应用提出了看法。
This paper discusses the problem of time warping in speech signal recognition from the perspective of maximum a posteriori estimation and discusses the relationship between the derived method and the commonly used dynamic time warping method. The application of speech recognition put forward the view.