For someone who does not speak French well, dealing with the authorities can quickly become difficult. In Togo, a programme aims to reduce this barrier by teaching digital tools to recognise several languages spoken in the country. The stakes go beyond technology: they concern access to public services and the place of Togolese languages in the digital economy.
Local languages in digital services
French is the official language of the Togolese administration, but it coexists with many languages used in everyday life. In this context, tools able to understand speech or text in local languages could make certain procedures easier, particularly for people less comfortable with written French.
The ‘Voix du Togo’ programme, led by the Ministry of Public Service Efficiency and Digital Transformation, plans to collect spoken and written data to train artificial intelligence models. This technology enables software to perform tasks such as recognising speech, translating text or generating a response.
Data collection begins with Ewe, Kabye and Moba. The programme is then expected to expand to other national languages. For each of the three pilot languages, the stated targets are to gather 50 hours of validated audio recordings, 6,000 pairs of translated sentences and contributions from at least 500 speakers. The data must be checked before being used to train the models.
Planned contributions include recording sentences, transcribing audio files and translating texts on an online platform. Speakers’ participation is crucial: their voices, words and usage can help the tools better reflect the ways people actually speak in Togo.
A project whose use still needs to be tested
Potential applications include public administration, healthcare and the justice system. In time, a user could complete certain procedures orally in a language they know. A patient could also describe their symptoms without first having to translate them into French. These are stated aims, not yet proof that such services work on a large scale.
Togo AI Lab, which works on artificial intelligence and data science projects, held public tests of the platform in Lomé on 28 September 2026. The aim of this stage was to gather user feedback and assess how well people understood its features. It comes before the stated uses can be considered available to the population.
This distinction matters. The quality of a voice tool depends on the diversity of the voices, accents and situations recorded, as well as the accuracy of transcriptions and translations. If the data mainly represent certain speakers or regions, the system may be less effective for others. The announced validation will therefore need to assess linguistic quality as well as the practical usefulness of the results.
Data protection is also an issue. A voice can sometimes identify a person. The arrangements for consent, storage, access and reuse of recordings will therefore affect participants’ trust. The public launch of the programme alone does not establish what specific safeguards will apply; this information will need to be clearly communicated to contributors.
Access beyond the capital
The potential benefits will not be the same for everyone. In formal services, a reliable tool could reduce difficulties in understanding when people deal with the authorities. For public officials, it could help direct requests, provided it does not replace human interaction when a situation is complex. In rural areas as well as urban neighbourhoods, access will also depend on the availability of a phone, an internet connection and suitable support.
The languages selected for the pilot give the project an initial scope, but they do not reflect the full linguistic diversity of Togo. The planned expansion to other languages will raise questions about resources: gathering quality data, mobilising speakers and specialists, and then keeping the tools up to date. A three-year programme provides a framework, but does not in itself guarantee coverage of all languages or equal service quality.
The project is part of a broader effort to develop local capacity in artificial intelligence. Togo AI Lab also works in areas such as agriculture, healthcare, education and the environment. The aim is to develop skills and tools rooted in the country’s needs, rather than relying solely on technologies designed for other languages and contexts.
Regional potential, with results still to prove
Across West Africa, where exchanges and administrative procedures span several linguistic areas, the methods for collecting and validating data could be of interest to other initiatives. But languages and their uses cannot automatically be transferred from one country to another. Each project will need to draw on the speakers and specialists concerned.
For Togo, the next measure of progress will not be just the volume of data collected. It will also be important to see whether the models correctly understand speakers with varied profiles, whether errors are identified and corrected, and whether the planned services actually become accessible beyond demonstrations in Lomé. It is this test that will show whether ‘Voix du Togo’ can help make public administration easier to understand and strengthen the presence of local languages in the digital sphere.



