Binaural recording using artificial heads is a common method to record acoustic scenes spatially. A drawback of this method has been its inability to incorporate head movements by listeners during playback, with detrimental effects such as in-head localization. Adapting recorded acoustic scenes to head movements has so far required different recording techniques involving microphone arrays and more than two channels. This contribution presents a new approach to adapt conventional two-channel binaural recordings to head movements by analyzing and modifying the cues contained in the binaural signal. According to a listening test, the proposed algorithm works with signals recorded in reverberant single-source as well as multi-source scenes. Therefore, it opens up the possibility to improve localization, externalization and realism of traditional binaural recordings.