Practised pre-processing of audio signals and understood the use of Mel-frequency cepstral coefficients (MFCC) features. Audio recordings of vowels of varying lengths and stress were recognised using DTW, a dynamic programming algorithm. A 16 desnsity Gaussian Mixture model was trained on MFCC features of audio recordings of digits (0 to 9) of class members. In testing phase, my own recordings were used. The argmax of the posterior probability tells to which digit the sample belongs