Musemuzi is a cutting-edge language tool designed to facilitate seamless communication between Kinyarwanda and English speakers. Leveraging the latest advancements in Artificial Intelligence, Musemuzi offers accurate and nuanced translations, empowering users to navigate diverse linguistic landscapes with confidence. Musemuzi utilizes advanced speech recognition technology to convert spoken Kinyarwanda or English into text, allowing for immediate translation. Our state-of-the-art AI engine then deploys sophisticated algorithms to deliver precise and contextually aware translations, ensuring your message transcends linguistic barriers.

Musemuzi is available for both Android and iOS (iPhone & iPad) platforms.

DeepKIN is a software toolkit for Kinyarwanda natural language processing tasks. The toolkit include open source code in Python and PyTorch and pre-trained models for various tasks. Based on our morphology model for Kinyarwanda, we devise morphology-based language models that adapts Transformer architecture to morphologically-rich languages.

MorphoKIN, the morphological analyzer and generator for Kinyarwanda is a software library for analyzing Kinyarwanda words for natural language processing applications.

KinyaTTS repository on GitHub is a codebase for Kinyarwanda speech synthesis (text-to-speech) based on MB-iSTFT-VITS2 model in PyTorch. The repository provides a pre-trained multi-speaker TTS model that can generate moderate quality Kinyarwanda output.

We developed a streaming speech recognition model that can be deployed on light-weight architectures such as smartphones without GPU. The model is based on Mamba state-space model architecture and is implemented in Pytorch. A pre-trained model is available for download fromGoogle Drive. The model can be exported into a compiled TorchScript Lite model deployable on smartphones (See DeepKIN package). An Android app that runs this model is available here. A compiled demo APK File is available for download at this Google Drive file.

Igisoro is a traditional board game of the Mancala family played primarily in Rwanda, Burundi, DR Congo and Uganda. We implemented a mobile version of the board game that can be played on Android or iOS platforms. The game features single plays as well as double plays where two people can play the game across two devices using BlueTooth Low Energy (BLE) wireless technology.

IGISORO is available for both Android and iOS (iPhone & iPad) platforms.