Natural Language Processing
Systems that understand, process and generate Somali text with high accuracy and cultural relevance.
Research
We research and build AI systems for Somali language understanding, generation, speech, translation and multimodal applications.
Research areas
Systems that understand, process and generate Somali text with high accuracy and cultural relevance.
Language models trained and fine-tuned on Somali data for reasoning, content generation and conversational AI.
Somali speech recognition (speech-to-text) and natural-sounding Somali speech synthesis (text-to-speech).
AI translation between Somali and other major languages, to help communication and access to knowledge.
Systems that combine text, speech, images and video, designed for Somali speakers and local use cases.
The datasets and benchmarks that every Somali model is trained and measured on.
Core projects
Showing 8 projects
A foundation language model trained on Somali data for reasoning, writing and conversation.
Accurate speech-to-text for spoken Somali.
Natural-sounding Somali voices generated from text.
Neural machine translation between Somali and English.
High-quality Somali datasets for training and evaluating AI systems.
An AI assistant that helps students and teachers learn in Somali.
A growing, structured collection of knowledge in the Somali language.
AI tools that help public institutions serve citizens in Somali.