Code and Technical Projects · 2023–present
LLM Text-to-Performance
A system that generates a script, converts it into audio, and delivers unrehearsed dialogue to performers line by line in real time.
- Role
- Design and code
- Technologies
- Python, Large language models, Text-to-speech, Audio segmentation, QLab
- Collaborators
- Asa Wember
A large language model generates a theatrical script from audience suggestions or predetermined parameters. The script is converted to audio, divided into individual lines, and transmitted to performers through earbuds.
The performers speak dialogue they have never rehearsed, listening to and embodying the text in the same moment. The audience witnesses both the content of the scene and the visible act of each performer receiving and processing new information in real time.