Isolated Telugu Speech Recognition using MFCC and Gamma tone features by Radial Basis Networks in Noisy Environment

In this paper, Radial basis neural networks[1][12][17] have been examined for speech recognition using speech features MFCC (Mel frequency Coefficients) and Gamma tone frequency coefficients for isolated Telugu words in noisy environment. Speech feature vectors are used to train, validate and test the Radial basis neural networks.Experiments conducted in Office environment under the presence of light/fans/Air conditioning noises, and the results are analyzed. MFCC (Mel frequency Cepstrum Coefficients) are preferably used features for Speech recognition in ASR Systems. In recent trends another speech features extraction technique called Gamma tone frequency Coefficients are being experimented for ASR. Both MFCC and Gamma tone features [2] [3] [4] have been analyzed under the same noisy conditions for isolated telugu words. The design and development of the neural network, features extraction are implemented in MATLAB environment [12][17] and analyzed the results.