| Summary: |
Recently, voice recognition technology is widely used to produce lip sync animation. However, the reference in current speech recognition system is not in Indonesian language, and therefore, the results in Indonesian system become inaccurate. To develop a better, suitable system, Indonesian vowels should be used in the speech recognition systems. Because there is a vowel in every syllable, vowels have an important role in lip sync animation. Hence, features of vowels need to be selected properly to increase accuracy of the system. This process can be performed by comparing mean of formant values of a vowel in a word at each position; the front, the middle, and the back. The results show different values for those three positions. If the values are significantly different, the formant values are unfit for vowels reference of voice recognition system. The result of this research shows that the formant values of vowel 'i' and 'o' are not different significantly in all positions. Formant values of vowel 'a' and 'u' do not differ significantly in only one position. And the values of formant of vowel 'e' are significantly different between each position.
|