Automatic scoring of pronunciation quality with hybrid measure

Bin Dong, Fengpei Ge, Fuping Pan, Shui Duen Chan

Research output: Chapter in book / Conference proceedingConference article published in proceeding or bookAcademic researchpeer-review

1 Citation (Scopus)

Abstract

In this paper, a hybrid measure for automatic scoring of Mandarin pronunciation quality is presented. Different to prevalent algorithms, mono-phone-based and tri-phone-based acoustic models are applied and two types of features are combined to get the score of "Goodness of Pronunciations" with Support Vector Machine algorithm, which are the average logarithm of the frame-based posterior probability and the normalized logarithm of the phoneme-based posterior probability. With the hybrid measure, the average correlation coefficient between machine scores from automatic system and the rater scores is improved from 0.8379 to 0.8549 which almost reach the coefficient 0.8564 between different raters.
Original languageEnglish
Title of host publication2009 International Conference on Artificial Intelligence and Computational Intelligence, AICI 2009
Pages381-384
Number of pages4
Volume3
DOIs
Publication statusPublished - 1 Dec 2009
Event2009 International Conference on Artificial Intelligence and Computational Intelligence, AICI 2009 - Shanghai, China
Duration: 7 Nov 20098 Nov 2009

Conference

Conference2009 International Conference on Artificial Intelligence and Computational Intelligence, AICI 2009
CountryChina
CityShanghai
Period7/11/098/11/09

Keywords

  • Automatic scoring
  • Hybrid meansure
  • Speech recognition

ASJC Scopus subject areas

  • Artificial Intelligence
  • Computational Theory and Mathematics
  • Computer Graphics and Computer-Aided Design
  • Software

Cite this