Research
Pioneering research into Generative AI for low-resource contexts.
We are pushing the boundaries of Generative AI applications to solve real problems and exploring innovative ways to apply them effectively in resource-constrained environments.
Our Focus Areas
We are collecting large amounts of multimodal datasets – text, audio, image and video – to build generative AI models that performs downstream tasks.
Text
We are developing text models that uses a large volume of under-represented language datasets meticulously augmented and codemixed. These models can perform downstream tasks such as Q& A in native dialects with greater accuracy.
Audio
We are exploring the application of AI to process voice data and audios in under-represented languages for Speech-to-Speech (STS), and Automatic speech Recognition (ASR)
Image
Our research centers on models that are able to generate highly contextual and representative images, from text descriptions, built using a vast database of Afrocentric visual images.
