Maximum F1-Score Criterion Based Discriminative Feature Compensation Training Algorithm for Automatic Mispronunciation Detection

HUANG Hao; XU Hai-hua; WANG Xian-hui; Wushour Silamu

doi:10.3969/j.issn.0372-2112.2015.07.007

您当前的位置：

首页 >

文章列表页 >

Maximum F1-Score Criterion Based Discriminative Feature Compensation Training Algorithm for Automatic Mispronunciation Detection

更新时间：2025-07-16

- Maximum F1-Score Criterion Based Discriminative Feature Compensation Training Algorithm for Automatic Mispronunciation Detection
- Acta Electronica Sinica Vol. 43, Issue 7, Pages: 1294-1299(2015)
- 作者机构：
  
  1. 新疆大学信息科学与工程学院,新疆,乌鲁木齐,830046
  2. 南洋理工大学Temasek实验室,新加坡,639798
  3. 新疆大学信息科学与工程学院,新疆,乌鲁木齐,830046
  4. 南洋理工大学Temasek实验室,新加坡,639798
- 作者简介：
- 基金信息：
  
  National Natural Science Foundation of China (No.61365005, No.60965002)
- DOI：10.3969/j.issn.0372-2112.2015.07.007
  CLC： TN912.34
- Published：2015
- 稿件说明：
移动端阅览
HUANG Hao, XU Hai-hua, WANG Xian-hui, et al. Maximum F1-Score Criterion Based Discriminative Feature Compensation Training Algorithm for Automatic Mispronunciation Detection[J]. Acta Electronica Sinica, 2015, 43(7): 1294-1299.
DOI：

HUANG Hao, XU Hai-hua, WANG Xian-hui, et al. Maximum F1-Score Criterion Based Discriminative Feature Compensation Training Algorithm for Automatic Mispronunciation Detection[J]. Acta Electronica Sinica, 2015, 43(7): 1294-1299. DOI： 10.3969/j.issn.0372-2112.2015.07.007.

摘要

为提高自动发音错误检测性能

提出一种区分性特征补偿训练算法.该方法将高斯后验概率矢量经过线性变换后作为偏移量补偿至传统的谱特征.将经过正确度标注的语音数据库上的发音错误检测

1值的最大化作为变换参数的训练准则.推导了目标函数对变换参数的偏导数公式

并利用无约束参数优化例程L-BFGS更新变换参数.发音错误检测实验表明该方法能够有效增大训练和测试集的

1值.并且训练和测试集的精确度、召回率也都有明显提高.在特征优化的基础上进行模型参数训练

检错性能较单独的区分性特征训练、单独的区分性模型训练都有进一步改进.

Abstract

To improve the performance of automatic mispronunciation detection

a discriminative feature compensation training algorithm is proposed.The method is to train a matrix projecting from posteriors of Gaussians to a normal size feature space

and then to add the projected features to traditional spectral features.The matrix is trained according to maximum

1-score criterion

which aims at maximizing the empirical mispronunciation detection

1-score on the annotated speech database.Mispronunciation detection experiments have shown the method is effective in increasing

1-score

precision and recall on both the training data and evaluation data.It is

also shown model parameter discriminative training on new features obtained further improvements over both model training and feature training.

关键词

Keywords

references

Views

1377

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

A Priori Information-Based Lightweight Convolutional Neural Network for Video Denoising

A Recording Device Identification Algorithm Based on Improved PNCC Feature and Two-Step Discriminative Training

Diffusion Features in Radar Specific Emitter Identification

A Discriminative Segmental Feature Transform Method Based on Uncorrelated Matching Pursuit

Related Author

MENG Xian-dong

ZHU Shu-yuan

ZHU Qiang

SHENTU Min-jian

LI Xin-chao

ZHU Zheng-yu

Alexander I Rudnicky

WANG Zhi-feng

Related Institution

School of Information and Communication Engineering， University of Electronic Science and Technology of China

Peng Cheng Laboratory

School of Electronic and Information Engineering, South China University of Technology

School of Computer Science, Carnegie Mellon University

School of Electronic and Information Engineering South China University of Technology Guangzhou Guangdong China

⁰