In this paper, wavelet transform technique and neural network is used for development of Speaker Verification System for short utterances. The sampled data undergo 4-level decomposition in wavelet decomposition technique. DCT (Discrete Cosine Transform) is performed on the dataset, to improve the features extraction process. This study includes Hilbert Transform, which shows the importance of magnitude and phase for speaker classification and their performance was shown. Hilbert Transform is explored, to analyze performance of phase for the data. The features are then, fed to feed-forward back propagation neural network for further classification. The proposed technique is evaluated on fixed phrase of the RedDots dataset and self-recorded numerical dataset. The proposed method performs effectively up to 95% recognition rate.