In music, a melodic motif is often played repeatedly in different pitch ranges and at different times. Event-related potential (ERP) studies have shown that the mismatch negativity (MMN) reflects memory trace processing that encodes two separate melodic lines (“voices”) with different motifs. Here we investigated whether a single motif presented in two voices is encoded as a single entity or two separate entities, and whether motifs overlapping in time impede or enhance encoding strength. Electroencephalogram (EEG) from 11 musically-trained participants was recorded while they passively listened to sequences of 5-note motifs where the 5th note either descended (standard) or ascended (deviant) relative to the previous note (20% deviant rate). Motifs were presented either in one pitch range, or alternated between two pitch ranges, creating an “upper” and a “lower” voice. Further, motifs were either temporally isolated (silence in between), or temporally concurrent with two tones overlapping. When motifs were temporally isolated, MMN amplitude in the one-pitch-range condition was similar to that in the two-pitch-range upper voice. In contrast, no MMN, but P3a, was observed in the two-pitch-range lower voice. When motifs were temporally concurrent and presented in two pitch ranges, MMN exhibited a more posterior distribution in the upper voice, but again, was absent in the lower voice. These results suggest that motifs presented in two separate voices are not encoded entirely independently, but hierarchically, causing asymmetry between the upper and lower voice encoding even when no simultaneous pitches are presented.