Most of the new generation mobile phones are equipped with smart and unique features that provide a great experience in music, internet, capturing videos and many more applications. But the same experience is not achieved in voice communication when the surrounding environment is too noisy. Hence development of algorithms for processing complex voice is very much essential for speedy and correct estimation of speech signals. This skill, demand processing of voice signal digitally for feature extraction and remove the undesired noise sources. In this paper, the degraded signal is pre processed before passing through Mel-filter bank further Mel frequency Cepstral coefficients are found and feature extraction is carried out. Since the voice signals tend to have variation in temporal rate, adaptive equalization is used for feature matching techniques. The test pattern in this paper presents an improvement in speech index by a factor of 20-30%.