Abstract:
Automatic Speech Recognition (ASR) is a very useful technology to human being. Eventhough this technology has been developed for many years it still could not be used widely for daily activities. One of the problem that limits its usage is the diffrerences between languages. Once an ASR is trained using a language, that ASR could not properly recognize speech in another language. Bahasa Indonesia is one of the most used language in the world. There are more than 250 millions peoples use Bahasa Indonesia everyday. But the availability of ASR in Bahasa Indonesia is still limited. There are several speech databases in Bahasa Indonesia which are developed by big companies but they could not be accessed publicly. To further help the development of ASR in Bahasa Indonesia a speech database in Bahasa Indonesia must be developed. In this paper, an attempt to collect sound of speech vowel in Bahasa Indonesia is reported. Vowel sound from several major etnics of Bahasa Indonesia user which have big differences in sound are recorded. The speakers are divided into two groups by gender (male or female) and age (adult). The frequency characteristic for each group are investigated and the similarities are measured using correlation to further improve the quality.
