Documenting the Path to Open Source: Community-Driven Development of an ASR System for Jamaican Patois (Patois.ai)

My project, Patois.ai, aims to address the linguistic exclusion of Jamaican Patois in the world of artificial intelligence and automatic speech recognition technology. This exclusion has profound implications for the justice, equity, and inclusion of Patois speakers in the digital world. By documenting the process of building an ASR system for Jamaican Patois and sharing it openly through the NVIDIA NeMo open-source community, Patois.ai contributes to the advancement of inclusive AI technology and paves the way for other marginalized languages to join the digital conversation. Additionally, the project raises important legal questions surrounding data collection, model training, and intellectual property rights, contributing to the vital dialogue on responsible AI development and governance. Dr. Shieber's expertise in computational linguistics, open access, and ethics aligns strongly with my project's aims. His research on ASR and ethical AI development can provide valuable guidance and collaboration opportunities.

• 276