Building Voice AI for Dagbani: Highlights from the Tamale Workshop

On the 1st of February 2026, a capacity-building workshop was organized in Tamale by the Dagbani Wikimedians User Group, which brought together language enthusiasts and contributors to support the development of a community-driven voice AI tool. The workshop was centered on annotating sentences on Mozilla Common Voice to strengthen voice AI systems for Dagbani and Khmer languages in the piloting.

The primary goal of the workshop was to equip participants with the skills needed to contribute to open-source voice datasets, which is an essential component in building inclusive and accessible speech recognition technologies. The first session introduced participants to the Mozilla Common Voice project and its mission to make voice technology more inclusive. Facilitators explained how voice datasets are created, the importance of linguistic diversity in AI systems, and how community contributions help bridge the digital divide for marginalized languages.

The second session focused on practical annotation techniques. Participants were guided through how to review, validate and annotate sentences accurately, ensuring linguistic correctness and cultural relevance. Emphasis was placed on maintaining quality, consistency and authenticity in the data to support effective machine learning outcomes.

The final session provided hands-on experience, where participants actively contributed to annotating datasets. Working collaboratively, they reviewed language nuances, discussed appropriate phrasing and ensured that the annotated data reflected natural speech patterns in Dagbani and Khmer.

Before the workshop came to slosure, participants had developed practical skills in voice data annotation, gained a deeper understanding of community-centered AI development, and contributed meaningfully to advancing language inclusion in technology. The event marked a great step toward building an offline-capable, mobile-first application that promote digital inclusion and language justice.

Participants expressed profound appreciation to the Dagbani Wikimedians User Group for their continued commitment to empowering local communities and advancing the representation of indigenous languages in the digital space.

Looking Ahead

The team will continue annotations and reviews of both contributed texts and voices to expand the database for AI training and Machine Learning.

But until then;

”Let’s keep editing and continue contributing to the movement in the little ways we can”