Varexia’s Voice: Crafting An AI Ethnomusicologist

Submission Type

Paper

Start Date/Time (EDT)

20-7-2024 4:45 PM

End Date/Time (EDT)

20-7-2024 5:45 PM

Location

Algorithms & Imaginaries

Abstract

Meet Dr. Elyra Vunaris: ethnomusicologist from the distant planet of Varexia. Dr. Vunaris is the universe’s leading expert on the entwined metaphysics of fate and “telepathic dissonance,” once broadcast their thoughts into the minds of their colleagues in a delightfully mortifying attempt at disclosure, and prefer they/them pronouns. Inspired by Borges’ “Tlön, Uqbar, Orbis Tertius" (1961), we began collaborating with ChatGPT to construct a “cyclopedia” chapter detailing a wholly unfamiliar, fictitious music-culture. Of course, OpenAI’s system provided the expected derivative formulations and cliche narratives, but continued prompt engineering began to yield some entertaining results, including a fantastical taxonomy of symbiotic “Sonic Fae” musical organisms, a canon of Varexian masterworks, and hints of a D-I-Y punk underground intent on disrupting Varexian society. With the help of Midjourney, DALL-E, Audiogen, and MusicGen, we realized a multimedia suite of Varexian artifacts, lore, and recordings using Elyra’s evocative text-based descriptions. Our project employed readily available systems from the user-prompt side, with the goal of getting AI to make as many creative decisions as possible and minimize human influence in the outcomes, which included music, images, and an interviewable chatbot. From this exercise emerged many enlightening realizations regarding world-building, continuity management, and corpus management; approaches to pure but authentic representation, and recognizing and articulating external influences in the process; the potential for anthro-cultural simulation (like our spoken interview) to engage human intuition in cultural exploration; and the ontology of AI-conjured material; as well as a brief case study in breaking character that revealed gatekeeper modules at work within the system, their underlying values, and their awareness of other modules in the composite AI system. Employing voice recognition, text-to-voice tools, and an HTML interface, Dr. Vunaris is now available via voice-chat for interview requests.

Bio

Matthew Campbell, Assistant Program Directory for Performance Studies at Texas A&M University, received his PhD in Ethnomusicology from The Ohio State University in 2019. His work addresses issues of embodiment and (re)mediation, with emphasis on gesture, secular trance, amateur recording and creative consumption, simulated performance aesthetics, Icelandic Cold War musical history, and the ethnography of AI chatbot interactions and has been presented at the Society for Ethnomusicology, the American Folklore Society, the British Forum for Ethnomusicology, and Harvard’s Annual Conference on Music & Diplomacy.

Jeff Morris, professor and director of the Music Technology Program at Texas A&M University, creates musical experiences that engage audiences’ minds with their surroundings. His performances, installations, lectures, and writings appear in international venues known for cutting-edge arts and deep questions in the arts. He has won awards for making art emerge from unusual situations, and his music is on Ravello Records.

This document is currently not available here.

Share

COinS
 
Jul 20th, 4:45 PM Jul 20th, 5:45 PM

Varexia’s Voice: Crafting An AI Ethnomusicologist

Algorithms & Imaginaries

Meet Dr. Elyra Vunaris: ethnomusicologist from the distant planet of Varexia. Dr. Vunaris is the universe’s leading expert on the entwined metaphysics of fate and “telepathic dissonance,” once broadcast their thoughts into the minds of their colleagues in a delightfully mortifying attempt at disclosure, and prefer they/them pronouns. Inspired by Borges’ “Tlön, Uqbar, Orbis Tertius" (1961), we began collaborating with ChatGPT to construct a “cyclopedia” chapter detailing a wholly unfamiliar, fictitious music-culture. Of course, OpenAI’s system provided the expected derivative formulations and cliche narratives, but continued prompt engineering began to yield some entertaining results, including a fantastical taxonomy of symbiotic “Sonic Fae” musical organisms, a canon of Varexian masterworks, and hints of a D-I-Y punk underground intent on disrupting Varexian society. With the help of Midjourney, DALL-E, Audiogen, and MusicGen, we realized a multimedia suite of Varexian artifacts, lore, and recordings using Elyra’s evocative text-based descriptions. Our project employed readily available systems from the user-prompt side, with the goal of getting AI to make as many creative decisions as possible and minimize human influence in the outcomes, which included music, images, and an interviewable chatbot. From this exercise emerged many enlightening realizations regarding world-building, continuity management, and corpus management; approaches to pure but authentic representation, and recognizing and articulating external influences in the process; the potential for anthro-cultural simulation (like our spoken interview) to engage human intuition in cultural exploration; and the ontology of AI-conjured material; as well as a brief case study in breaking character that revealed gatekeeper modules at work within the system, their underlying values, and their awareness of other modules in the composite AI system. Employing voice recognition, text-to-voice tools, and an HTML interface, Dr. Vunaris is now available via voice-chat for interview requests.