The control of relative timing is evident in many social activities including synchronisation in music performance. The Linear Phase Correction Model of musical ensemble (Wing et al., 2014) suggests each player in a string quartet corrects the timing of their next note in proportion to the asynchrony with fellow players on the previous note. To examine melodic and rhythmic influences on this correction in a live musical setting, 12 participants played a violin melody part in synchrony with a violin duet who played the same melody and an accompaniment pitched below the melody. In half of the trials, the duet’s melody was played live, and in the other half it was a recording, which was not made explicit to the participants. In both cases, the accompaniment was always a recording. The timing structure of the duet was either simple (both parts in 2/4 time), or complex (melody in 2/4 and accompaniment in 6/8 time). After each trial, participants rated the perceived influence of melody and accompaniment on the timing of their playing. Timing performance, measured using note acoustic onsets, revealed higher correction gains in the linear phase correction model with the melody than with the accompaniment, and the effect was more apparent in the complex timing condition. Complementing this finding, ratings of perceived influence indicated greater influence of the melody than the accompaniment, especially in the simple timing condition. There was no effect of whether the melody was live or recorded. These patterns of correction and perceived influence indicate players take correction cues from the score part that is more similar to their own in terms of melody and rhythmic structure.