Isolated Spoken Word Identification in Malayalam using Mel-frequency Cepstral Coefficients and K-means clustering

This paper proposes an approach to recognize isolated spoken Malayalam words. The paper deals with a speech feature extraction technique based on MFCC and K-mean clustering. We used six Malayalam words for the experiment and hundred speakers are used to identify these words in the testing phase. The words and stored in a database and later identified. MFCCs are calculated in both training and testing phase for different words, once in a training session and once in a testing session later. The word is identified according to the minimum quantization distance which is calculated between features of each word in training phase and the individual word in testing phase. This system is proposed for real time, speaker- independent word recognition systems with limited number of words.