ABSTRACT
Movement of the lips and tongue during speech is an important component of facial animation to increase the immersion of users in digital contents. Mouth movement during speech is ongoing and relatively rapid, and the movement encompasses a number of visually distinct positions. Also, the shape of mouth must be synchronized to the dialogue. Since the most of energy and time duration of speech signals are accumulated in vowels, it is required to recognize vowels from the dialogue of dubbing artists to produce animation key for the lip shape of virtual characters. Previous results of researches related to speech recognition, especially vowel recognitions, can be applied for synchronizing the mouth movement with a dialogue in digital contents such as animations and e-learning contents (Hwhang et el. 2013a). However, the principle techniques were mainly originated form
the English-speaking world and the Japanese. Since the vocal sound system of Korean is different from other languages, it should be careful that the existing or previous technique is directly applied for recognizing Korean vowels. Furthermore, most of previous research results related to Korean language were derived using the digital signal processing technique which treats Korean-speaking as an input sound signal without considering the characteristic of the Korean vocal sound system. There is possibility to improve the recognition algorithm and rate though the Korean phonemes always have the same phonetic description or value.
