Speech processing for isolated Marathi word recognition using MFCC and DTW features

Abstract- Even though speech recognition is a broad subject, the commercial and personal use implementations are rare. Although some promising solutions are available for speech synthesis and recognition, most of them are tuned to English. The acoustic and language model for these systems are for English language. Most of them require a lot of configuration before they can be used. If we go to the rural area across the India then still there are people who don’t understand English as well as can’t speak proper English. So this available speech recognition system is of no use for these people. There are so many regional languages in India, but being Maharashtrian we got inspired to think about Speech Recognition system for Marathi. There are several problems that need to be solved before speech recognition can become more useful. The amount of pattern matching and feature extraction techniques is large and the decision on which ones to use is debatable. After studying the history of speech recognition we found that the very popular feature extraction technique Mel frequency cepstral coefficients (MFCC) is used in many speech recognition applications and one of the most popular pattern matching techniques in speaker dependent speech recognition is Dynamic time warping (DTW). The signal processing techniques, MFCC and DTW are explained and discussed in detail and these techniques have been implemented in MATLAB