Inclusive Education through AI: A Framework for Swahili - English Audio Access in Tanzanian Libraries

Tanzania faces a critical shortage of accessible library resources for blind learners, despite 8.2 million people experiencing vision loss and 290,000 being blind. Census data show 3.6 million persons with disabilities (8%), who face disproportionate exclusion, including 48% illiteracy and only 3.1% employment. Libraries lack audio reading programs, and assistive technologies are scarce and costly. This paper proposes an AI-driven web framework integrating CNN-based OCR, LLM-powered text processing, and neural text-to-speech for Swahili and English. The solution enhances library accessibility, reduces dependence on proprietary tools, and promotes inclusive education in line with Tanzania’s disability rights commitments.