Feature selection using Fisher's ratio technique for automatic speech recognition

摘要

Automatic Speech Recognition involves mainly two steps; feature extractionand classification . Mel Frequency Cepstral Coefficient is used as one of theprominent feature extraction techniques in ASR. Usually, the set of all 12 MFCCcoefficients is used as the feature vector in the classification step. But thequestion is whether the same or improved classification accuracy can beachieved by using a subset of 12 MFCC as feature vector. In this paper,Fisher's ratio technique is used for selecting a subset of 12 MFCC coefficientsthat contribute more in discriminating a pattern. The selected coefficients areused in classification with Hidden Markov Model algorithm. The classificationaccuracies that we get by using 12 coefficients and by using the selectedcoefficients are compared.