Smart Braille Translator: Integrating Braille-to-Speech and Speech-to-Braille Using Raspberry Pi

In the Philippines, approximately half a million individuals are estimated to have visual impairments, with a significant number lacking the ability to read Braille. This study addresses various challenges in the development of a Braille translation system, such as ensuring accurate conversion of Braille characters into text and speech. The system leverages a Raspberry $\operatorname{Pi} 4$ as a microcontroller, employing a Convolutional Neural Network (CNN) for scanning Braille characters, followed by Natural Language Processing (NLP) and Braille mapping for translation. Moreover, the study will integrate the principles of assistive technology theory throughout the development of the system. During data gathering, the proponents discovered that most respondents from the Philippine Blind Union community had only a basic understanding of Braille. The study ultimately achieved its three primary objectives by implementing the system successfully. Regarding Braille conversion, the system demonstrated a $\mathbf{1 0 0 \%}$ accuracy rate for simple dot patterns and $\mathbf{9 8 \%}$ for more complex patterns. Additionally, the camera scanning yielded $96 \%$ accuracy in well-lit conditions and 70% accuracy in dim lighting. The final system evaluation, based on ISO 25010:2023 standards, showed positive feedback from respondents who tested the system.