计算机科学 ›› 2010, Vol. 37 ›› Issue (1): 214-216.

• 人工智能 • 上一篇    下一篇

基于正交实验设计的语音识别特征参数优化

韩志艳,王健,王旭   

  1. (渤海大学信息科学与工程学院 锦州121000);(东北大学信息科学与工程学院 沈阳110004)
  • 出版日期:2018-12-01 发布日期:2018-12-01
  • 基金资助:
    本文受国家自然科学基金((50477015)资助。

Parameter Optimization Based on Orthogonal Test Design in Speech Recognition System

HAN Zhi-yan,WANG Jian,WANG Xu   

  • Online:2018-12-01 Published:2018-12-01

摘要: 鉴于语音识别性能与所选用的语音特征参数密切相关,提出一种系统性的实用的特征参数优化方法—基于方差的正交实验设计法。首先进行因素(语音特征参数)和水平的选择,再根据数理统计与正交性原理,从大量的实验点中挑选适量的具有代表性、典型性的点构造正交表进行正交实验,最后通过计算对正交实验结果进行分析,找出最优的特征参数组合。与目前参数的简单组合方案相比较,新方法的误识率下降了5.6%,响应时间减少了181. 37ms。实验结果表明,正交实验设计用于语音特征参数优化是有效的,对后续研究具有指导意义。

关键词: 正交实验设计,语音识别,参数优化,正交性

Abstract: The performance of speech recognition depends on the choice of speech parameters. A systematic and practical method of the parameters selection based on the variance orthogonal test design was proposed. Firstly,chose factors (speech parameters) and levels. And then according to the principle of mathematical statistics and orthogonality, picked out proper and representative points from a large number experimental points to construct orthogonal table. Finally, calculated and analysed the experimental results, and the optimal set of process parameter values was discovered. Moreover, the word error rate is reduced 5. 6% and response time is reduced 181. 37ms when compared with that of the traditional parameter selection method. The experiment results show that the application of orthogonal test design is effective in parameter optimisation. The conclusion provides guidance for the subsequent research.

Key words: Orthogonal test design, Speech recognition, Parameter optimization, Orthogonafity

No related articles found!
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!