Malay vowel

Keywords— Formant Frequency, Fundamental Frequency, Malay Vowel, Gender Identification in Children, Children Speech. 

One of the most important components of the human speech is formant frequency and has been studied in children and adults across different languages so far. As matter of the fact, formant frequencies make vowels identifiable for listener. Because of the different formants position in vowels of male and female, formant frequency can be used in gender identification especially in the first three formant frequencies of vowels [1, 2]. A need to look into the Malay children s formant and fundamental frequencies of Malay vowels, is obvious and the effect of them on the matter of gender identification is considerable. The aim of the present study is to find any systematic differences related to gender in fundamental and formant frequency values of vowel /a/ which is produced by Malay children between 7 to 12 years old. 

Keywords— Malay Vowel Recognition, Time Delay Neural Network, Automatic Speech Recognition, Children Speech. 

In this study, TDNN used six output units instead of three in the original TDNN. Each of the output units corresponds to each of the Malay vowels. 

In this study, the vowel samples data were collected from Malay children between 7 to 12 years old. The sampling rate was set at 20 kHz which fulfilled the minimum requirements as suggested by previous studies [15]. Goldwave was used to collect and store the sound samples. The sustained vowel sounds were recorded from a total of 180 male children and 180 female children. The speech recording was done in quiet room. The subjects were asked to sit with body upright and mouth facing the microphone. Each of the Malay vowels was sustained for more than 5 seconds. 

The speaker independent vowel recognition using Time Delay Neural Network was presented. The result showed that TDNN was capable of discriminate different Malay vowels. The temporal characteristic of the speech was learned properly by the recognition model. Accuracy of 81.92% suggested that the model is suitable for vowel recognition for children. 

Ting H N and Yunus J (2004) Speaker-Independent Malay vowels recognition of children using multi layer perception. IEEE regional conference TENCON, 2004, pp 68-71... 

The study investigated the first four formant frequencies and fundamental frequency of sustained vowel /a/ in normal Malaysian Malay children between 7 to 12 and applied them for gender identifying. The results showed that the second formant frequency was the best and also fundamental frequency played a role in the matter of gender identification. The combination of FO and F2 also compared with other possible combination to get a good accuracy. The best accuracy belongs to the F2 and combination of FO and F2. 

Speaker-Independent Vowel Recognition for Malay Children Using Time-Delay... 

Fig. 2 Time delay neural network for Malay children vowel recognition...

---