Some common words (name, address, E-mail ID, ... ), we press daily and we are habituated to press it in same rhythm, which is unique and can be used to segregate and distinguish people. In this paper we are considering rhythm not only the entered common words or some sequence of common characters. Here machine intelligence relies on the fact that it stores the typing style of some daily used words which are supported by the user and can be used as a secret key. Recognising typing style promises a parameter like behavioural biometric characteristics that may facilitate non-intrusive, cost effective and continuous monitoring. But this technique, as of now, suffers from accuracy level and performance. In order to realize, this technique in practice a higher level of security and performance together with low cost version is needed with an error to an accepted level. Hence, it is highly essential to identify the controlling parameters and optimise the accuracy and performance as well as cost with new algorithms.