Technology news around the ecosystem!

Ubini Is Bringing Edo Language into the Age of Artificial Intelligence

The future of artificial intelligence in Africa may depend on how well technology understands the continent’s indigenous languages. While global AI companies continue investing heavily in language models, thousands of African languages remain underrepresented in the digital resources used to train these systems. Nigerian startup Ubini is attempting to address this gap by transforming Edo language, oral histories and cultural knowledge into structured digital data that machines can understand.

Founded in 2024 by Stephanie Ohumu, a journalist and artificial intelligence expert, Ubini is building a digital archive of Edo conversations, traditional knowledge, recipes and cultural experiences. Its mission goes beyond language translation. The company wants to preserve knowledge that could disappear with older generations while creating resources for future African-language technologies.

The idea emerged from a personal experience. Following the death of her grandmother in 2023, Ohumu realised that recipes, expressions and cultural knowledge she had previously relied on her grandmother to explain were no longer easily accessible. That experience highlighted how much Edo heritage remained undocumented.

Addressing the African-language AI gap

Africa is home to more than 2,000 languages, but relatively few have sufficient digital datasets to support modern natural language processing. This creates a significant challenge for AI developers attempting to build systems that accurately understand African languages.

Edo presents additional technical difficulties because it is a tonal language. The placement of tonal markings can change the meaning of words, making accurate transcription essential.

Without reliable training material, AI systems can generate translations that sound convincing but contain incorrect expressions or meanings. Such errors can undermine confidence in digital language tools, particularly among people who cannot independently verify the translations.

Research published in the Proceedings of Machine Learning Research in 2026 examined support for African languages in large and small language models. It identified only 41 supported African languages and 23 publicly available datasets, highlighting the considerable gap in language representation.

Ubini sees this shortage as both a cultural preservation challenge and a technology development opportunity.

Turning oral conversations into machine-readable data

Ubini’s approach involves recording conversations with older Edo speakers, particularly individuals above 70 who possess knowledge of traditional practices, community history and everyday expressions.

The company has interviewed approximately 20 people and collected more than 20 hours of audio recordings. Its research covers areas such as traditional food, family histories, ceremonial expressions and cultural practices.

However, collecting recordings is only the beginning. The more difficult task is converting spoken Edo into accurate written material.

The team manually transcribes conversations and translates them into English. Each initial transcription undergoes quality checks by two professors at the University of Benin, helping ensure that tonal markings, spelling and meanings are accurately represented.

The verified material is then organised into XML files. These structured files contain speaker information, Edo-language transcriptions, English translations and recording details.

This process is important because artificial intelligence systems require organised and reliable datasets to recognise language patterns, understand context and generate meaningful responses.

Ubini’s immediate target is to collect 50 hours of audio by December 2026. It also aims to produce 150 hours of transcribed and translated material by the end of 2027, before expanding its archive towards 250 hours by 2029.

Beyond artificial intelligence: Building a cultural business

While language technology is central to Ubini’s mission, Ohumu sees opportunities beyond developing datasets for AI companies.

The startup believes Edo cultural knowledge can become valuable intellectual property for the creative industry, including documentaries, films, books and educational content.

One example is Mia Mia, a documentary featuring 85-year-old Edo speaker Maria Evbagharu Osazee. The film explores cultural changes and the importance of preserving knowledge across generations.

The documentary premiered at the Uncolonial Film Festival in Germany in August 2026 and was subsequently selected for the Human Screen Festival in Tunisia.

Ubini intends to develop similar creative projects while building its language archive, allowing cultural preservation and commercial storytelling to grow together.

Its proposed revenue model includes licensing language datasets to technology developers and generating income from media productions and intellectual property partnerships.

However, the company remains bootstrapped and faces considerable operational expenses, particularly around transcription, translation and quality assurance. Ohumu is therefore seeking patient funding to expand the team and accelerate data collection without compromising the project’s long-term objectives.

Why Ubini matters for Africa’s technology future

Ubini represents an emerging opportunity within Africa’s artificial intelligence ecosystem: developing technology around local languages rather than expecting indigenous languages to adapt to systems designed primarily for global languages.

Reliable Edo datasets could eventually support voice-enabled banking, digital education, customer service and other applications for people who prefer communicating in their native language.

For Nigerian fintech companies and technology developers, this could create opportunities to reach customers who face language barriers when using digital services.

Nevertheless, scaling the project will require sustained investment, linguistic expertise, community participation and strong data-governance practices. Protecting contributors’ consent and ownership interests will also be important as cultural materials become commercial datasets and creative products.

Ubini is demonstrating that Africa’s AI opportunity is not limited to building applications or adopting foreign technology. It also involves creating the foundational data that makes intelligent systems relevant to African communities.

By turning oral conversations into structured digital records, the startup is attempting to preserve Edo’s cultural heritage while preparing the language for a future in which machines can understand and communicate with its speakers.

Sources and further reading

Leave a Reply

Your email address will not be published. Required fields are marked *