Speech data for African languages.

South African speech data collection and code-switched datasets — helping train the next generation of local-language AI models. Built for AI teams and research partners.

What's included

Where we help.

South African speech data collection

Code-switched speech datasets

Data annotation & quality assurance (coming soon)

AI dataset project management

Capacity & specs

What we can deliver.

Capacity

100–500 hours of code-switched speech data per project.

Language pairs

Zulu–English, Sotho–English, Pedi–English, Tswana–English.

Delivery format

Standard annotated format, with custom formats available on request.

Typical timeline

30–60 days from scope confirmation to delivery.

Want to see a sample before scoping a project? We have demos available on request.

How it works

Our process.

Scope & requirements

We align on languages, dialects, use case and volume needed for the dataset.

Collection & annotation

We run structured data collection with quality checks built into the process.

Delivery & QA

Datasets are reviewed, packaged and delivered within 30–60 days, in the format your team needs.

Get a proposal

Tell us about your AI speech data needs.

We'll come back to you with a scoped proposal — mention it below if you'd like a sample demo first.

No payment · No obligation · We'll respond within 1 business day

Strategy & Growth PR & Communications Brand Development Digital Presence AI Data Services Creative & Organisational