This study aims to replicate the irrelevant speech effect (ISE) in a local context and, more important, is the first to directly investigate if musical information can reduce impairments imposed by the ISE on a serial word recall task. Thirty-five undergraduates from the National University of Singapore performed serial recall on 10 word lists. The lists were presented under 5 auditory conditions, namely: Music-Only, Combined (music with background speech), Scrambled music with background speech, Background Speech-Only and White Noise conditions. The Scrambled condition contained the same piece of music as the Combined condition except that it was re-arranged in a random fashion; the mission of this condition was to specifically provide a comparison basis to test if “musical structure” per se actually attenuates the ISE. A significant main effect of music conditions emerged. ISE was successfully replicated, where a significantly lower percentage of correct words was recalled in the Background Speech-Only condition compared to all other conditions. ISE was also successfully attenuated, but the present data suggest that musical structure per se was not (at least not entirely) responsible for the attenuation, since the Scrambled condition had superior performance than both the Combined and Background Speech-Only conditions. Here, we propose and discuss several novel theoretical models involving changing acoustical features, selective attention, and arousal to account for the present findings.