Know a resource we should list?
Especially community-built work that search engines bury.
Build with Igbo
Building with Igbo? Start with what’s already here.
The starting point is not zero. Researchers and communities have already built Igbo datasets, speech recordings, models, tools and research. The hard part is often finding them and knowing where to start.
For companies, that means Igbo speech technology does not have to be built from scratch. Pick what you are building below, or browse all resources.
What are you trying to build?
Need Igbo speakers to evaluate your product, or help choosing a starting point? Talk to TokataHQ
Browse all resources
The full directory
Search by name or keyword, or filter by kind of resource and review status.
resources categories primary listings checkedDirectory edition:
Each entry records the source we checked and what we could confirm about access and licensing. Entries marked “Needs review” are leads we could not fully confirm. Listing a resource is not an endorsement of its accuracy, safety, licensing or upkeep.
For product teams
Where to start if you are adding Igbo
- Check the tracker first. Find the capability you are building and see what is documented, what we have tested and what still needs testing. Open the tracker
- Use the speech data that already exists. The directory includes Igbo speech listings such as NaijaVoices and the Nigerian Common Voice Dataset. Review each dataset’s Igbo coverage, licensing, and fit for your use case.
- Evaluate against an Igbo-specific benchmark, not an African-language average. An average across many languages can hide how a model performs on Igbo.
- Get a fluent Igbo speaker to review the output. Tone and diacritics change meaning. A model that reads fluently to a non-speaker can still be wrong.
- Talk to us. TokataHQ can connect you to the research, the data, and the people. Talk to TokataHQ
Need fluent-speaker evaluation?
Tell us the product, the capability, and the type of review you need.