Many animals communicate using composite “multi-modal” signals, which simultaneously address different senses. Components can be redundant, with each prompting similar response by receivers, or non-redundant, with each prompting different responses. Furthermore, the composite signal may have a similar meaning to one or both components, or a novel meaning. We used robot models to test the role of visual and vocal signals in the multi-modal duets of Australian magpie-larks, Grallina cyanoleuca. Pairs produce coordinated antiphonal vocal duets, which function to defend the territory, but they also coordinate these vocal duets with visual duets, which usually entail alternating wing movements. However, there has been no study of this visual component of display. We describe these multi-modal signals, and measure aggressive responses by territorial pairs to each component independently, as well as to the multi-modal signal. We assessed territorial aggression by the number of songs and flights. Each sex had a repertoire of both song types and wing movements. Partners rarely used the same song type within a duet, yet almost always used the same type of wing movement. Birds responded aggressively and equally to multi-modal and purely vocal duets, but less aggressively to purely visual duets, so that the visual component did not increase the perceived territorial threat. However, surprisingly, birds gave more alarm calls to multi-modal duets than to either vocal or visual duets. Our results imply that the vocal duet conveys the message to neighbours about territoriality, while the visual component might just attract attention. However, the coordination of the type of wing movement within displays, independent of song type, indicates that the birds must look at their partner during multi-modal duets. We therefore suggest that the visual display may have a role in coordinating vocal duets within pairs, analogous to a conductor’s baton, and so have a different audience.