Google Is Expanding WAXAL Beyond 21 Languages — What It Means for African Researchers

Why It Matters

The move accelerates AI accessibility for African languages and creates a sustainable talent pipeline, positioning the continent as an active contributor to global AI research.

Key Takeaways

• WAXL expands to 27 African languages, more upcoming
• Google partners with Masakhane Hub for community‑led development
• TranslateGemma model enables local fine‑tuning of dialects
• AI Residency replaced by scalable Student Researcher program
• Google’s AI Center in Accra supports continent‑wide upskilling

Pulse Analysis

The launch of WAXAL marked a pivotal step in addressing the chronic data shortage that has hampered speech‑recognition research for African languages. By releasing a large‑scale, openly licensed corpus covering 21 initial languages—and now extending to 27—the initiative supplies the raw material needed for training robust models. This expansion not only improves voice‑assistant accuracy across the continent but also invites global developers to experiment with under‑represented linguistic data, fostering a more inclusive AI landscape.

Google’s collaboration with the Masakhane African Languages Hub signals a strategic shift toward community ownership. Rather than a top‑down rollout, Masakhane’s grassroots network ensures that language resources are curated by native speakers, preserving dialectal nuances often lost in generic datasets. The accompanying TranslateGemma model serves as a high‑performance foundation that African researchers can fine‑tune, accelerating the creation of localized translation and transcription tools. By hosting new language releases on Hugging Face, Google encourages open‑source contributions, turning the dataset into a living repository that evolves with community input.

Talent development is equally central to Google’s African AI agenda. The sunset of the traditional AI Residency program gave way to a scalable Student Researcher model, allowing universities across the continent to place top‑tier students directly into cutting‑edge projects. Coupled with the AI Community Center in Accra, which offers workshops, mentorship, and collaborative space, these efforts build a pipeline of skilled engineers who can both consume and produce AI innovations. As Google continues to integrate AI into projects like Open Buildings and food‑security forecasting, the expanded WAXAL dataset and its supporting ecosystem position Africa as a proactive player in the next wave of global AI advancement.

Google is expanding WAXAL beyond 21 languages — What it means for African researchers

By Victoria Fakiya · Senior Writer · February 9, 2026

Last week, Google announced the launch of WAXAL, a large‑scale open speech dataset for 21 Sub‑Saharan African languages, such as Hausa, Yoruba, Igbo, Luganda, Swahili, and Acholi.

The initiative is the brainchild of a partnership between the tech giant and African research institutions, and it’s designed to enhance artificial‑intelligence tools for African languages.

To better understand the initiative and what Africa stands to benefit from it, Abdoulaye Diack, Program Manager at Google Research, shares more insights with Techpoint Africa on the progress made so far and how African researchers can position themselves to be part of it.

After launching WAXAL, which covers 21 Sub‑Saharan African languages. Are there plans to expand to other parts of Africa and cover more languages?

Yes. While WAXAL provides a critical foundation for 21 Sub‑Saharan languages, we are already expanding our reach. We are currently at 27 languages, with several more in the pipeline.

Our roadmap, as our SVP James Manyika highlighted, is now to collaborate directly with the Masakhane African Languages Hub. This partnership is a shift from just “adding numbers” to building a sustainable, community‑led ecosystem.

Masakhane is a grassroots initiative that ensures African languages are developed by and for Africans. For a glimpse of what’s coming next, we recommend watching our Hugging Face repository, the project’s technical heartbeat, where new languages are being processed and released.

How can other African researchers interested in assisting with translation join in adding their local languages to the mix?

We are focused on ensuring that AI isn’t just something that ‘happens’ to Africa, but something that African researchers can lead and define. To support that, we’ve provided resources like TranslateGemma, an open model that serves as a high‑performance foundation.

While it already supports languages like Hausa, Yoruba, and Igbo, the real value lies in the hands of local researchers who can take this tool and fine‑tune it for their specific dialects and linguistic nuances.

There is a 2‑year AI residency programme that trains young African researchers. Is it still operational, and are there plans to expand to other African countries?

The landscape of how we nurture talent has evolved. While the original AI Residency Program has been sunset, we’re still committed to the next generation.

We’ve shifted our focus toward the Student Researcher roles. The beauty of this model is its scale—it isn’t restricted to one or two countries; we are actively hiring top‑tier students from across the entire continent.

In fact, we’re already looking ahead to our 2026 intake. We want to catch that talent early, while they are still in their academic journey, and bring them into the fold of global‑scale research.

Google Research and its impact in Africa

People are often surprised to learn that Google has been conducting world‑class research right here in Africa for years. Whether it’s the Open Buildings project, which has mapped over 500 million buildings across the continent to support urban planning, or work on food security and weather forecasting, the impact is real and global.

Our AI Community Center in Accra is a testament to that; it’s a hub for upskilling and collaboration. These technologies aren’t just abstract concepts, but tools that are changing lives on the ground.