Starmirror Training Performance
Starmirror Ensemble supported by Byrdland

Starmirror ensemble: Sara Gouzy, Marina Kerdraon-Dammekens, Andria Prokopa

On several days throughout the exhibition Starmirror, participants are invited into a live process of AI training, as the hall of KW transforms into a recording studio where local choirs, an ensembles, and visitors contribute their voices in call-and-response sessions.

During these sessions, participants are asked to sing hymns from Herndon’s and Dryhurst’s newly created songbook, generated by a model trained on Ordo Virtutum (1151)—a morality play by Hildegard von Bingen (1098–1179), the German Benedictine abbess and polymath, in which a soul must choose between good and evil. The artists took inspiration from Hildegard von Bingen, seeing analogies between her visions of a celestial order between humans and the cosmos, and the protocols shaping today’s cultural and technological infrastructures.

Each Sunday, a Berlin-based choir joins the ensemble for the call-and-response sessions, singing alongside visitors. To join the sessions, no prior knowledge or preparation are required, just openness and enthusiasm for collective singing. They are training performances, offering a direct, participatory experience with the human contributions that define AI.

The recordings of the sessions form the basis of a public choral dataset, which will train an AI “Berlin choir” to debut at Kunstsammlung Nordrhein-Westfalen (K21) in Düsseldorf.

The vocal ensemble Byrdland is an eight-member solo-voice ensemble from Berlin. The group combines the joy of exploring new horizons and emotions together with a lively dialogue with the audience. A special focus lies on the connection between Renaissance and contemporary vocal music, creating soundscapes that surprise and move – from Palestrina to pop, from Byrd to Birdland.

Usage Rights

By taking part in the Starmirror training performances, participants grant the artists and KW Institute for Contemporary Art unlimited usage rights to the voice recordings they contribute. This includes:

• Record, store, and reproduce the recordings.
• Use the recordings for Artificial Intelligence (AI) training and modification.
• Present and distribute the recordings, both in their original form and as AI-generated works (e.g. choir songs).
• Make recordings and AI outputs publicly accessible, including online and for promotional purposes.
• Edit, adapt, or combine the recordings with other material as needed for the creation and presentation of the project.
• Recordings may also be released as part of the public domain.

Participation is voluntary and without financial compensation. Due to the large number of contributors, participants will not be individually credited in connection with the resulting works.