Voix du Togo: the plan to bring public services into local languages 🇹🇬🤖

Cliquez ici pour lire en français

In Togo, artificial intelligence could soon make it easier for citizens to deal with the administration in their own national languages. Under a pilot program called “Voix du Togo” (“Voice of Togo”), the Ministry of Public Service Efficiency and Digital Transformation, backed by the United Nations Development Programme (UNDP), plans to build AI models that can process Kabiyè, Ewe and Moba.

The starting point is a simple problem. For part of the population, dealing with the administration remains hard when information and services come in a language they don’t master well. French is the language of the administration, while many Togolese speak other languages at home. The project aims to narrow that gap with technology that better understands the languages people actually speak every day.

Three national languages first 💻

In its initial phase, “Voix du Togo” plans to train AI models on three national languages: Kabiyè, Ewe and Moba. The ambitions go further, though. Authorities say they eventually want to bring around fifty national languages together on a single platform open to all users.

If it works, citizens could gradually get administrative information in a language they understand better. A farmer in the Kara region, in northern Togo, could one day ask a public service a question in Kabiyè and receive an answer in the same language. For now, that is a goal of the program, not a live service.

The aim is to make public services more accessible to people whose mother tongue differs from the language used in administrative exchanges.

Citizens as trainers of the models 🎙️

To learn to recognize and process national languages, AI models need reliable linguistic data. The program therefore plans to recruit speakers directly, through a platform built to collect voice recordings and text.

In practice, participants register with their phone number and state which languages they speak and write. They then complete tasks such as reading texts aloud, transcribing recordings, translating sentences and checking other users’ contributions.

Each accepted contribution is paid. The mechanism makes citizens part of building the datasets the models need, while compensating them for their work.

Linguists and the communities concerned are also expected to check the quality of what is collected. The idea is to limit the errors and approximations that could affect how AI systems understand these languages.

Reaching people with limited connectivity 🌍

Online contributions alone won’t be enough. Some residents have no smartphone or no stable internet connection. Without another channel, their languages and usages risk being underrepresented in the data used to train the models.

To work around this, field agents are expected to visit communities directly and collect contributions from residents. The goal is to widen participation beyond people already comfortable with digital tools.

In this way, the program says it wants to better reflect the country’s linguistic diversity, rural areas included. The aim is data that reflects how people really speak, rather than only how they speak in big cities.

Language inclusion at the heart of digital transformation 🏛️

Beyond the technology, “Voix du Togo” is about inclusion. The program aims to ease access to public services, but also to healthcare and justice, by letting citizens use the languages they speak daily.

According to UNDP, the tools could also have applications in administration and entrepreneurship. By taking national languages into account, AI could help widen access to services and digital information.

Still, the initiative’s reach will depend on the quality of the data collected, how well it handles linguistic particularities, and whether new languages can be added progressively. With “Voix du Togo”, the country is betting on AI to bring digital services closer to the linguistic realities of its people.

💬 Do you think AI can really make public services easier to access across Africa by working in national languages? Tell us in the comments.