Nowadays, wavelet has been widely applied in extracting features of the signal for automatic speech recognition system. Wavelets have many families that are determined by their mother function and order. The use of different wavelets to analyze the same signal would bring different results. In many cases, a trial and error procedure is used to select the optimal wavelet family. That is because there are no particular wavelet basis functions that can be applied to the entire speech signals. Therefore, it is necessary to analyze the similarity between the speech signal and the wavelet base function. One of the methods that can be used is cross-correlation. In this study, the degree of correlation is determined between wavelet base function and Indonesian vowels. The influence of gender and consistencies of the results are also used in the analysis. The results show that db45 and db44 are most similar to male and female vowels utterance, respectively. For consistencies, only vowel e gives a consistent result. Overall, db44 is most similar to all Indonesian vowels utterance.
Real Time Impact Factor:
Pending
Author Name: Syahroni Hidayat, Habib Ratu P. N., Danang Tejo Kumoro
URL: View PDF
Keywords: automatic speech recognition; cross-correlation; Indonesian vowels; wavelet; wavelet basis function determination
ISSN: 1411-8289
EISSN: 2527-9955
EOI/DOI: http://dx.doi.org/10.14203/jet
Add Citation
Views: 1
