A Discriminative Segmental Feature Transform Method Based on Uncorrelated Matching Pursuit

CHEN Bin; NIU Tong; ZHANG Lian-hai; QU Dan; LI Bi-cheng

doi:10.3969/j.issn.0372-2112.2016.12.016

您当前的位置：

首页 >

文章列表页 >

A Discriminative Segmental Feature Transform Method Based on Uncorrelated Matching Pursuit

更新时间：2025-07-16

- A Discriminative Segmental Feature Transform Method Based on Uncorrelated Matching Pursuit
- Acta Electronica Sinica Vol. 44, Issue 12, Pages: 2924-2931(2016)
- 作者机构：
  
  1. 解放军信息工程大学信息系统工程学院,河南,郑州,450001
  2. 西南电子电信技术研究所上海分所,上海,200434
  3. 解放军信息工程大学信息系统工程学院,河南,郑州,450001
  4. 西南电子电信技术研究所上海分所,上海,200434
- 作者简介：
- 基金信息：
- DOI：10.3969/j.issn.0372-2112.2016.12.016
  CLC： TN912
- Published Online：25 December 2016，
  
  Published：2016
- 稿件说明：
移动端阅览
CHEN Bin, NIU Tong, ZHANG Lian-hai, et al. A Discriminative Segmental Feature Transform Method Based on Uncorrelated Matching Pursuit[J]. Acta Electronica Sinica, 2016, 44(12): 2924-2931.
DOI：

CHEN Bin, NIU Tong, ZHANG Lian-hai, et al. A Discriminative Segmental Feature Transform Method Based on Uncorrelated Matching Pursuit[J]. Acta Electronica Sinica, 2016, 44(12): 2924-2931. DOI： 10.3969/j.issn.0372-2112.2016.12.016.

摘要

为了提高基于分帧特征变换方法的稳定性，提出了一种基于分段的区分性特征变换方法.该方法将特征变换当成高维信号的稀疏逼近问题，采用状态绑定的方法训练得到基于域划分的线性变换矩阵（Region Dependent Linear Transform，RDLT）和基于最小音素错误准则均值补偿的特征（mean-offset feature Minimum Phone Error，m-fMPE）变换矩阵，将两者的特征变换矩阵构成过完备的字典；采用强制对齐的方式对语音信号进行分段，以似然度最大化作为目标函数，利用匹配追踪算法对目标函数迭代优化，自动地确定各语音信号段中的变换矩阵及其系数.为保证特征变换的稳定性，在选择变换矩阵过程中引入相关度测量，去除相关的特征基矢量.实验结果表明，相比于传统的RDLT方法，当声学模型分别采用最大似然和区分性准则训练时，识别性能分别可以提高1.63%和2.23%.该方法同时能应用于语音增强和模型区分性训练中.

Abstract

A discriminative segmental feature transform method is proposed to promote the stability of the frame based method.The feature transform is considered as the sparse high dimensional approximation problem.Firstly

a set of feature transform matrices are estimated by tied-state based training of RDLT (Region Dependent Linear Transform) and m-fMPE (mean-offset feature Minimum Phone Error)

and the transform matrices are integrated into an over-complete dictionary.Then

the speech signal is segmented through force alignment.Finally

following the matching pursuit to optimize the likelihood objective function iteratively

the transform matrices of each segment are selected from the dictionary and the corresponding coefficients are automatic determined in the optimization process.Further

to guarantee the stability of the transform matrices

a correlation measurement is introduced to remove the correlated basis in the recurrence process.The experimental results show that

compared with the traditional RDLT method

when the acoustic model is trained with maximum likelihood and discriminative training criterion separately

the recognition performance can be improved by 1.63% and 2.23% respectively.The method can also be applied to speech enhancement and model discriminative training.

关键词

Keywords

references

Views

322

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

An In-Vehicle Interaction Speech Enhancement and Recognition Method Based on Lightweight Models in Complex Environment

Suppression Method of the Interference Sound Sources by Estimated Steering Vector Based on the Focusing Signal Subspace

An End-to-End Chinese Speech Recognition Algorithm Integrating Language Model

An Over-complete Learned Dictionary-Based Image De-noising Method

Related Author

LIAN Xiao-yu

XIA Nan

DAI Gao-le

YANG Hong-qin

ZHOU Jing

BAO Chang-chun

ZHANG Xu

LÜ Kun-ru

Related Institution

School of Information Science and Engineering， Dalian Ploytechnic University

Speech and Audio Signal Processing Laboratory, Faculty of Information Technology, Beijing University of Technology

Ministry of Education Key Laboratory of Symbol Computation and Knowledge Engineering， Jilin University

School of Computer Science， Zhuhai College of Science and Technology

College of Computer Science and Technology， Jilin University

⁰