We study collaborative aggregation of structured sequences, using musical harmonization as a concrete and well-defined testbed. While our motivating application is music, the broader goal of this work is to understand how collective inputs over structured strings can be aggregated algorithmically in a principled way–an issue that also arises in text aggregation and related domains. Concretely, given multiple harmonization suggestions for a fixed melody, we design and analyze aggregation algorithms that aim to balance two competing objectives: (i) representing the agents’ suggestions faithfully, and (ii) producing a sequence that is internally coherent according to a deliberately simple transition-based notion of harmony. We introduce a formal model, propose several aggregation rules inspired by social choice, and study their computational properties. Our experimental evaluation shows that relatively simple methods, in particular plurality- and Kemeny-based rules, perform consistently well under both representation and coherence criteria.