Ultrafast vibrational spectroscopy experiments, together with molecular-level theoretical interpretation, can provide important information about the structure and dynamics of complex condensed phase systems, including liquids. The theoretical challenge is to calculate the instantaneous vibrational frequencies of a molecule in contact with a molecular environment, accurately and quickly, and to this end a number of different methods have been developed. In this paper we critically analyze these different methods by comparing their results to accurate benchmark calculations on azide/water clusters. We also propose an optimized quantum mechanics/molecular mechanics method, which for this problem is superior to the other methods.