I Am Sitting in a (Latent) Room

Nicholas Shaheed, and Ge Wang

Proceedings of the International Conference on New Interfaces for Musical Expression

Abstract:

In this paper we describe I Am Sitting in a (Latent) Room, a real-time structured group improvisation system inspired by Alvin Lucier's "I Am Sitting in a Room," and the general process of degrading sound by repeatedly passing it through an acoustic medium. But there is a twist. Unlike "I Am Sitting in a Room," which unfolds as a gradual process with no further interaction once the process has begun, I Am Sitting in a (Latent) Room gives the improvisers the ability to intervene and interact with the process of degradation in real time. An audio clip is repeatedly encoded and decoded through two parallel instances of a bespoke variational autoencoder (VAE) model. On top of this process, the performers manipulates the model's latent embeddings in real-time, exploring the latent space (or "room") of the model over the course of the performance. Two performances with the composer and live-coding duo RGGTRN are presented. This work explores human-in-the-loop AI systems through group improvisation, interactive AI performance, and creating datasets as a part of the compositional process.

Citation:

Nicholas Shaheed, and Ge Wang. 2024. I Am Sitting in a (Latent) Room. Proceedings of the International Conference on New Interfaces for Musical Expression. DOI: 10.5281/zenodo.13904872

BibTeX Entry:

  @article{nime2024_49,
 abstract = {In this paper we describe I Am Sitting in a (Latent) Room, a real-time structured group improvisation system inspired by Alvin Lucier's "I Am Sitting in a Room," and the general process of degrading sound by repeatedly passing it through an acoustic medium. But there is a twist. Unlike "I Am Sitting in a Room," which  unfolds as a gradual process with no further interaction once the process has begun, I Am Sitting in a (Latent) Room gives the improvisers the ability to intervene and interact with the process of degradation in real time. An audio clip is repeatedly encoded and decoded through two parallel instances of a bespoke variational autoencoder (VAE) model. On top of this process, the performers manipulates the model's latent embeddings in real-time, exploring the latent space (or "room") of the model over the course of the performance. Two performances with the composer and live-coding duo RGGTRN are presented. This work explores human-in-the-loop AI systems through group improvisation, interactive AI performance, and creating datasets as a part of the compositional process.},
 address = {Utrecht, Netherlands},
 articleno = {49},
 author = {Nicholas Shaheed and Ge Wang},
 booktitle = {Proceedings of the International Conference on New Interfaces for Musical Expression},
 doi = {10.5281/zenodo.13904872},
 editor = {S M Astrid Bin and Courtney N. Reed},
 issn = {2220-4806},
 month = {September},
 numpages = {6},
 pages = {333--338},
 presentation-video = {https://youtu.be/BfasOUklu7I?si=KisN4odmQE0CcxcB},
 title = {I Am Sitting in a (Latent) Room},
 track = {Papers},
 url = {http://nime.org/proceedings/2024/nime2024_49.pdf},
 year = {2024}
}