Bridging communication between hearing and visually impaired individuals via speech recognition and synthesis

<title>Abstract</title> <p>Effective communication remains a profound challenge for individuals with sensory disabilities, often leading to social isolation and limited opportunities for societal integration. To address these barriers, this paper presents a Speech Recognition and Synthesis Tool (SRST) designed to facilitate bidirectional communication specifically between hearing and visually impaired individuals. Built as a software-only solution, the system requires no specialized external hardware beyond a standard computer, ensuring high accessibility, cost-effectiveness and operational efficiency. This study was focused on acoustic signal processing via pattern recognition techniques for isolated words and short, trained conversational sentences, bypassing higher-level multimodal linguistic structures such as grammar, semantics, and pragmatics. Similarly, the text-to-speech component operates on rule-based algorithmic and dictionary-driven approaches to synthesize text into audible speech without contextual semantic analysis. By engineering a dual-window Graphical User Interface (GUI) within the MATLAB environment, a functional communication bridge was established. The system robustly transcribes real-time acoustic input into high-contrast digital text for hearing-impaired users while concurrently converting textual string inputs into synthesized, intelligible verbal waveforms via an optimized diphone concatenation engine for visually impaired individuals. Experimental validation indicates that the prototype meets its design objectives, achieving an average operational efficiency of approximately 90%. The architectural design of the system offers straightforward scalability, allowing the core vocabulary and phonetic database to be expanded using the established parameterization and modeling protocols. The prototype developed in this research demonstrates a baseline framework for language-independent speech processing, serving as a foundational architecture for more advanced speech-driven interfaces. To build upon the results of this study, we recommend integrating a machine translation layer between the ASR and TTS pipelines would allow the system to function as an automated language translator, expanding its utility to cross-lingual assistive communication.</p>