Human identification using unobtrusive methods is a challenging problem that has many applications in surveillance tasks. In this work we propose a set of biometric features extracted from a footstep audio signal that can be used to identify a person. Instead of using short-time spectral domain, Teager-Kaiser energy operator is employed to transform a time-domain signal into a representational domain where the intrinsic properties of the walking style of each individual become apparent. We show that these features are biometrically significant as they are physical correlates of human gait. The experimental results obtained using a recently recorded publicly available database show prominent results in a human identification task.