Voice conversion systems aim at finding a transformation function using statistical models. Mimicry/ Voice imitation is a natural voice transformation technique which sounds convincing to the listeners. It thus seems advisable to study the transformation used by human beings who perform mimicry. The objective of this study is to examine the various components of speech that are modified during voice imitation. To transform a given speech utterance to sound like that of a target utterance, the process needs to be understood at both production and perception level. In this paper the importance of source and system parameters and also the significance of different components of speech that contribute to the perception of imitation are studied. A flexible analysis-synthesis tool is used to modify the features of natural utterance and convert it to imitated utterance. Perceptual studies are carried out to understand if the modified features contribute to imitation. The results show that a combination of features is varied by the imitator to achieve imitation and they vary depending on the target speaker.