A new method for mispronunciation detection using Support Vector Machine based on Pronunciation Space Models

Wei, Si; Hu, Guoping; Hu, Yu; Wang, Ren-Hua

doi:10.1016/j.specom.2009.03.004

Cited by 85 publications

(66 citation statements)

References 21 publications

(26 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…They think that a complete active learning process is divided into four stages: task definition, goal setting and planning, strategy execution and meta cognitive regulation " [13]. Task stage is the process of clarifying the new learning tasks according to their own knowledge system and analyzing the relevant conditions and factors.…”

Section: A Active Learning Model Of Winne and Butlermentioning

confidence: 99%

Research and Software Development of Active Learning Mode for College Students Based on Computer Network

2016

Int. J. Emerg. Technol. Learn.

View full text Add to dashboard Cite

Abstract-With the development of information technology, especially the fast development of Internet technology, the network technology has expanded to all walks of life. Using Internet to acquire knowledge has become an important means for us. The paper sums up the current status of foreign active learning modes and the problems existing in computer network environment by referring to the relevant literature. What is more, it also further clarifies the basic concepts and connotations of active learning in the computer network environment and active learning itself by researching on its relevant theoretical basis and influential factors as well. "A new active learning mode in computer network environment" is constructed based on the analysis of the relevant factors of active learning in the computer network environment, taking college students as the studying object and the college English course as an example. According to this model, the active learning platform of college English network is designed and developed. The research conclusion is not only theoretically based, but also founded on the concrete practice and the application. Therefore, it is endowed with maneuverability so as to provide a research example for the general teachers.Index Terms-Internet technology; active learning mode; computer network environment; platform of college English

show abstract

Section: A Active Learning Model Of Winne and Butlermentioning

confidence: 99%

Research and Software Development of Active Learning Mode for College Students Based on Computer Network

2016

Int. J. Emerg. Technol. Learn.

View full text Add to dashboard Cite

show abstract

“…al [13] extracted acoustic-phonetic features and applied linear discriminant analysis (LDA), while Wei et. al [14] considered log-likelihood ratios (LLR) between the canonical phone model and a set of pronunciation variation models, and used support vector machine (SVM) for classification. Another direction is to explicitly model the possible mispronunciations based on linguistic knowledge.…”

Section: Related Workmentioning

confidence: 99%

“…As a result, some systems make use of training data from nonnative speech to improve system performance [11,14,17], although often it is not easy to get a good amount of well-labeled nonnative data. Most recently, Qian et.…”

Section: Related Workmentioning

confidence: 99%

Mispronunciation detection via dynamic time warping on deep belief network-based posteriorgrams

Lee

Zhang

Glass

2013

2013 IEEE International Conference on Acoustics, Speech and Signal Processing

View full text Add to dashboard Cite

In this paper, we explore the use of deep belief network (DBN) posteriorgrams as input to our previously proposed comparison-based system for detecting word-level mispronunciation. The system works by aligning a nonnative utterance with at least one native utterance and extracting features that describe the degree of mis-alignment from the aligned path and the distance matrix. We report system performance under different DBN training scenarios: pre-training and fine-tuning with either native data only or both native and nonnative data. Experimental results have shown that by substituting the system input from MFCC or Gaussian posteriorgrams obtained in a fully unsupervised manner to DBN posteriorgrams, the system performance can be improved by at least 10.4% relatively. Moreover, the system performance remains steady when only 30% of the annotations being used.

show abstract

“…An early example is Pols et al (1973), in which the automatic classification of Dutch monophthongs was investigated. More recently, research specifically targeted toward automatic pronunciation quality measures that can be employed in ASR-based CAPT systems has focused on confidence scoring (Witt, 1999;Franco et al, 2000;Yoon et al, 2010;Wei et al, 2009;van Doremalen et al, 2009) using ASR-based techniques. This type of research has shown that pronunciation errors can be accurately detected to a certain extent (Witt, 1999;Franco et al, 2000;Cucchiarini et al, 2009;Wei et al, 2009) and that difficulties may arise when it comes to identifying pronunciation errors that are based on subtle acoustic differences .…”

Section: Introductionmentioning

confidence: 99%

Automatic pronunciation error detection in non-native speech: The case of vowel errors in Dutch

Doremalen

Cucchiarini

Strik

2013

The Journal of the Acoustical Society of America

View full text Add to dashboard Cite

This research is aimed at analyzing and improving automatic pronunciation error detection in a second language. Dutch vowels spoken by adult non-native learners of Dutch are used as a test case. A first study on Dutch pronunciation by L2 learners with different L1s revealed that vowel pronunciation errors are relatively frequent and often concern subtle acoustic differences between the realization and the target sound. In a second study automatic pronunciation error detection experiments were conducted to compare existing measures to a metric that takes account of the error patterns observed to capture relevant acoustic differences. The results of the two studies do indeed show that error patterns bear information that can be usefully employed in weighted automatic measures of pronunciation quality. In addition, it appears that combining such a weighted metric with existing measures improves the equal error rate by 6.1 percentage points from 0.297, for the Goodness of Pronunciation (GOP) algorithm, to 0.236. V

show abstract

A new method for mispronunciation detection using Support Vector Machine based on Pronunciation Space Models

Cited by 85 publications

References 21 publications

Research and Software Development of Active Learning Mode for College Students Based on Computer Network

Research and Software Development of Active Learning Mode for College Students Based on Computer Network

Mispronunciation detection via dynamic time warping on deep belief network-based posteriorgrams

Automatic pronunciation error detection in non-native speech: The case of vowel errors in Dutch

Contact Info

Product

Resources

About