Subtractive Clustering Based Feature Enhancement for Isolated Malay Speech Recognition

This paper proposes a new hybrid method named SCFE-PNN, which integrates effective subtractive clustering based features enhancement and probabilistic neural network (PNN) classifier, had been introduced for isolated Malay word recognition. The proposed method of subtractive clustering features weighting is used as a data preprocessing tool, which designs at diminishing the divergence in features of the Malay word dataset, in order to further improve the recognition accuracy of the PNN classifier based speaker-dependent and speaker-independent mode. In this study, the melfrequency cepstral coefficients (MFCCs) were extracted from the selected Malay word. The experimental results show the effectiveness of the proposed SCFE technique. The proposed method shows promising average results of 99.61% (Speaker Dependent) and 96.21% (Speaker Independent) in distinguishing between the selected Malay words.