Feature extraction and classification of the Indonesian syllables using Discrete Wavelet Transform and statistical features

The major problem of most speech recognition systems is their unsatisfactory effectiveness (impact to recognition rate), efficiency (feature vector dimension), shift variance, and robustness in noisy condition. Feature extraction plays a very important role in the speech recognition process, because a better feature is good for improving the recognition rate. This paper presents a speech feature extraction by combining Discrete Wavelet Transform (DWT) and statistical method for recognizing the syllables sound in the Indonesian language. Three different mother wavelet transforms combined with statistical method (DWT-Statistical) are used as a feature extraction method. Multi-layer perceptron is used as a classifier after feature extraction process. This research aims to find the best properties in effectiveness and efficiency on performing feature extraction of each syllable sound to be applied in the speech recognition method on the intelligent systems. Experiments, in this study, show that the proposed method which uses 29 features set is applicable for feature extraction and classification of the Indonesian syllable. The results show that the average of recognition rate for the DWT-Statistical at the 7th level decomposition by using mother wavelet of Haar, Daubechies 2, and Coiflet 2 are 57.77%, 71.11%, and 67.77%, respectively.