The conversion of creep compliance to dynamic moduli is one of the promising ways to overcome the limitation of the dynamic measurements for wide range of frequency. Remarkable algorithms were developed by Evans et al. [Phys. Rev. E 80, 012501 (2009)] and Kim et al. [J. Rheol. 59, 237-252 (2015)]. Although the former is independent of any model of creep compliance, it suffers from the instability for experimental error. On the other hand, although the latter is a stable algorithm based on the model of Havriliak and Negami [Polymer 8, 161-210 (1967)], it has a problem of parameter identification. This paper is an effort to improve the approach of Kim et al. [J. Rheol. 59, 237-252 (2015)] by suggesting new approximate functions for the Laplace transform of creep compliance. We compared our algorithm with the two previous ones. (C) 2016 The Society of Rheology.