DuoConvo – AI-Powered Multilingual Communication Mobile Application

DuoConvo was born from a simple problem: language barriers make communication difficult, especially in multilingual communities like Ghana, where people speak different local and international languages. Many existing translation tools offer limited support for African languages, making everyday conversations harder for people who do not share a common language.

To address this challenge, I developed DuoConvo, an AI-powered mobile application designed to make multilingual communication more accessible. Built using React Native (Expo), Express.js, and MySQL, the application supports languages including English, Twi, Ga, Ewe, French, and Hausa. It combines speech recognition, multilingual sentence embeddings, FAISS-based semantic search, and text-to-speech technology to process spoken input, identify relevant translations, and deliver spoken responses. An LLM-based fallback helps handle conversations that cannot be matched confidently with existing translations.

The result was a functional mobile application that brought together mobile development, backend engineering, and artificial intelligence into one system. Beyond its technical implementation, DuoConvo demonstrated how AI can be applied to real-world communication challenges while giving greater attention to underrepresented African languages. The project strengthened my understanding of multilingual AI systems and reinforced my interest in building technology that solves meaningful problems.