作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

›› 2004, Vol. 30 ›› Issue (02): 159-161.

• 论文 • 上一篇    下一篇

基于长时频谱统计参数的说话人识别

于保华,吴亚栋,李治柱   

  1. 上海交通大学计算机系,上海交通大学计算机系,上海交通大学计算机系 上海 200030
  • 发布日期:2004-01-25

Speaker Recognition Based on Long-time Spectrum Statistical Parameters

YU Baohua,WU Yadong,LI Zhizhu   

  1. (Computer Science Department, Shanghai Jiaotong University, Shanghai 200030)
  • Published:2004-01-25

摘要: 把频谱特征的统计参数应用于与文本无关的说话人识别系统中,与LPC参数一起作为一种更有效的参数。在与文本无关的说话人识别中,长时频谱特征的统计参数与特定人发音特征相关而与文本无关的特性弥补了LPC参数与音韵相关的不足。

关键词: 长时频谱, LPC, 说话人识别

Abstract: This paper applies the long-time spectrum statistical parameters to the text-independent speaker recognition system, as more valid parameters with LPC together. In the text-independent speaker recognition system, the long-time spectrum statistical parameters which have close relation with the speaker and have little relation with the text make up for the shortcoming of the LPC parameters which have relation with the text.

中图分类号: