Towards Embodied AI

摘要

Opera has historically been a site for developing new and emerging technologies. The explosion of interest in AI and machine learning tools has created numerous possibilities for generating creative content in music, poetry, and imagery, and opera is an ideal testing ground for exploring how these tools might contribute to novel forms of creative expression in a Western context. This chapter outlines our efforts to explore the creative possibilities of embodied AI through a series of dedicated workshops and performances with an interdisciplinary research team. The goal of the ongoing Robot Opera project is to integrate diverse AI tools and explore their potential for machine performers. Our investigations focus on how emerging technologies challenge conventional understandings of voice, embodiment, and performance. We also explore how to design virtuosic and lively robot performers that not only imitate human voices and musicality but can generate their own unique sounds and musical expression. Following a brief history of robots in opera, we summarise our prior work with robot performance. We then describe our iterative design approach and summarise key insights from the workshops and performances, considering the trade-offs of working with existing robots and the possibilities for working with open, configurable systems in performance.