Improvement of speech signal extraction method using detection filter of energy spectrum entropy

Chung, Kyungyong; Oh, Sang-Yeob

doi:10.1007/s10586-015-0429-9

Cited by 10 publications

(2 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…3) Endpoint detection by improved spectral entropy method [15]. 4) Pitch extraction by vowel main body extension method.…”

Section: Extract Wa-language Syllable Classification Patternsmentioning

confidence: 99%

See 1 more Smart Citation

Wa Language Syllables Classification by Support Vector Machine Based on Genetic Algorithm

Fu¹,

Su²,

Pan³

2017

dtcse

View full text Add to dashboard Cite

Abstract. In this paper, propose the method to classify Wa syllables by support vector machine (SVM) based on genetic algorithm (GA). Firstly, select all the Wa syllables from the speech corpus recorded. Secondly, extract pitch and formant frequencies by vowel main body extension method and linear predictive coding method respectively. Thirdly, describe every syllable with a 12-dimensional feature vector composed of its mean values, minima and maxima, belonging to its pitch frequency sequence and its first three formant frequency sequences respectively. Fourthly, divide the obtained data set into training set and test set, the former use to train SVM, and the latter test SVM. Fifthly, normalize data set into [0,1], and select the radial basis function as kernel function. Finally, use the K crossover validation method (K-CV) based on genetic algorithm (GA) to get the best parameter values train SVM. As a result, the predictive accuracy of Wa syllables classification by test set arrives at 84.83%, which has shown that the proposed method is effective and feasible on Wa syllables classification.

show abstract

“…3) Endpoint detection by improved spectral entropy method [15]. 4) Pitch extraction by vowel main body extension method.…”

Section: Extract Wa-language Syllable Classification Patternsmentioning

confidence: 99%

“…4) Pitch extraction by vowel main body extension method. Formant extraction by linear predictive coding (LPC) [15] method based on adaptive variational mode decomposition (AVMD) [16].…”

Section: Extract Wa-language Syllable Classification Patternsmentioning

confidence: 99%