MusIAug : Audience Experience of Visual Augmentations for Human-AI Co-Improvisations

Human-AI co-improvisation on digital musical instruments opens many expressive opportunities. However, it is not yet clear what the audience perceives and understands of such performances, for example regarding the respective contributions of musicians and artificial agents. In this paper, we investigate the use of mixed-reality visual augmentations to represent the interactions between a pianist and an artificial musical agent, with the aim of enriching the audience experience. We present the implementation of the system and a study involving 23 participants and augmentations with different levels of detail. Our results suggest that visual augmentations could increase the confidence of audience members in what they perceive of the respective contributions of human and artificial agents and the dialogue between them. They also provide insights on preferences for personalised levels of detail in the augmentations. We believe that this research will help inform the design of co-improvisation systems to benefit the audience.

Audience Experience of Visual Augmentations for Human-AI Co-Improvisations Florent Berthaut, Ken Déguernel, Mathieu Giraud, Barbara Dang, Matteo Plichon, Olivier Capra AIMC 2026 - AI Music Creativity, 2026, Berlin, Germany