8. Echoes: Machine Whisper
24 Fall, Harvard University. With Viola Tan, Kida Huang.
-
Marcel Duchamp
"The creative act is not performed by the artist alone; the spectator brings the work in contact with the external world by deciphering and interpreting its inner qualifications."
Echoes is a system inspired by Alvin Lucier's I Am Sitting in a Room, which transforms sound through iterative re-recordings until words dissolve into pure resonance. Similarly, Echoes explores the randomness and interpretive fluidity within AI by iteratively generating music. Starting with a user-provided music clip, the system employs a music captioning model to describe the piece. This description, along with the original audio, is then fed into MusicGen to create the next music clip. Each cycle accumulates layers of interpretation, introducing subtle shifts and transformations that evolve over time.
Beyond audio, Echoes extends its creative process into the visual realm. The generated music captions are converted into prompts for Stable Diffusion, producing animations that harmonize with the evolving soundscape. This integration of sound and visuals results in a continuously transforming multisensory experience, allowing users to witness the iterative interplay of randomness, interpretation, and accumulation.
The project not only serves as a tool for automated music and animation generation but also invites reflection on the nature of AI creativity. By amplifying the randomness inherent in machine learning models, Echoes questions the boundaries of authorship, intention, and artistic evolution in the age of intelligent systems. It offers a poetic exploration of how meaning, form, and resonance emerge through iterative processes.