Descriptions of the mini projects: Responsible AI international community to reduce bias in AI music generation and analysis – https://music-rai.github.io/#mini-projects
(Summary via Generative AI): The event highlighted three innovative artistic mini-projects aimed at addressing bias in mainstream AI models by creating music in often overlooked or marginalized genres. The projects used low-resource AI models and small datasets to explore Responsible AI (RAI) themes.
Selected Artists and Projects:
- Fá Maria (aka HAUT):
- Project: “Erasure”
- Focus: A sound installation that explored the relationship between voice, gender identity, and AI technology. It used vocal datasets from underrepresented queer and trans voices to address biases in mainstream voice-generative AI models.
- Junson Park:
- Project: An experimental sound arrangement that combined traditional Korean music with modern electronic influences.
- Focus: Overcoming limitations in mainstream AI audio models by representing cultural nuances in traditional Korean music.
- Shreya Gupta:
- Project: “Jugalbandi: Call and Response Between Tabla Player and Drum Player”
- Focus: Developing a rule-based method to deconstruct complex rhythmic structures for AI models, aiming to make them more inclusive and culturally aware.
These projects aimed to highlight the challenges of bias in AI and demonstrated how RAI techniques could address them.
The event was moderated by: Nick Bryan-Kinns | About | University of the Arts London staff research profiles