Comparing Support Vector Machine and K Nearest Neighbor Algorithms in Classifying Speech

Speech classification identifies different speech patterns and is used in various applications such as speech recognition, speaker identification, and emotion recognition. In this paper, a comparison of the performance of two classification algorithms (Support Vector Machine (SVM) and K Nearest Neighbor (k-NN)) for classifying the American English language was made. The Texas Instrument and Massachusetts Institute of Technology (TIMIT) training and testing speech files were used in this study. Both the SVM and k-NN classifiers were built in MATLAB by parsing the phonemes from the TIMIT training speech files and generating the corresponding Mel Frequency Cepstrum Coefficients (MFCCs) for each phoneme. The MFCCs were used to build both classifiers. The built Classifiers were tested using files from the TIMIT speech test database to determine accuracy, precision, and recall for each built classifier algorithm. Results from the experiments showed that the SVM classifier outperforms the k-NN classifier in terms of precision and recall for all the TIMIT speech phonemes used in this study. Both SVM and k-NN performed equally as they relate to accuracy for the phonemes tested. These results suggest that the SVM classifier is a better choice for speech phoneme classification for precision and recall performance criteria.