VOICE & SPEECH
Speech data
Explore African speech data for AI training, with dedicated routes for Nigerian English and Yorùbá.
Explore African speech data →
Nigerian English speech data →
Yorùbá speech data →
Data types hub
Find the African AI training data your project needs.
Explore BSG DataWorks’ current speech-data pages, then see the commissioning routes for language, image, video, annotation, collection, and governance work. If your precise brief is not listed, start with the data type and intended model use—we will scope the right route with you.
Browse data types
Choose the data modality, then move to the right project route.
Some routes lead to a detailed page you can evaluate now. Others begin with a commission brief because the data is collected and prepared to the project specification rather than offered as a generic catalogue item.
LANGUAGE & TEXT
Nigerian language data
Commission text and language-data work around the languages, domains, use cases, and quality requirements relevant to your model.
IMAGE
Image data
Commission image collection and preparation around your object classes, environments, contributor profile, devices, and annotation needs.
VIDEO
Video data
Commission video-data work around the actions, settings, capture conditions, consent workflow, and quality criteria your model requires.
ANNOTATION
Annotation and labelling
Make existing text, image, video, or speech data more useful with agreed taxonomies, guidelines, quality checks, and delivery formats.
GOVERNANCE
Collection and compliance support
Plan a collection programme and make its operational evidence easier for procurement, governance, and buyer review teams to assess.
African data collection services →
AI training data compliance support →
Project routes
Data is only one part of a usable training-data project.
Move from the modality you need into the operational support that makes the project easier to specify, collect, review, and deliver.
Commission new data
Define languages, markets, contributor profile, devices, formats, volume, and quality criteria for a new African data-collection programme.
Prepare existing data
Turn existing source material into model-ready datasets with an agreed label schema, guidelines, quality-assurance process, and delivery structure.
Document the evidence trail
Bring consent, provenance, safeguards, and diligence questions into view before a buyer, procurement team, or governance reviewer needs them.
License a suitable corpus
Where an existing dataset is appropriate, tell us the model use, markets, volume, language, and licensing context so we can discuss the available route.
Start with a usable brief
The details that help us direct you to the right data route.
You do not need a fully written specification to begin. A few practical details will help us decide whether a published page, a new collection programme, annotation work, or a governance discussion is the best next step.
What the model needs
- Data type and intended model or product use
- Languages, domains, scenes, tasks, or participant profile
- Whether you need new data, existing data, or both
- Key exclusions or safety requirements
How the data should work
- Target volume and preferred file or metadata format
- Annotation or transcription requirements
- Quality criteria and evaluation needs
- Desired delivery route and project timeline
What needs to be reviewable
- Rights, consent, governance, or compliance expectations
- Buyer, vendor, or due-diligence questions
- Expected processing and delivery locations
- Any potential sensitive-data or re-identification concerns
What happens next
- We review the information you have
- We identify the relevant published or commissioned route
- We clarify the missing operational questions together
- We scope an appropriate next conversation
From discovery to scope
A simple route from “what data?” to “what next?”
This hub is designed to help you find a sensible starting point without asking you to fit a complex AI-data requirement into a generic product catalogue.
Choose
Start with the closest data modality: speech, language, image, video, or an existing dataset that needs preparation.
Explore
Read the relevant published page where one exists, or use the commissioning route if your project is specific to a market or model.
Brief
Share the requirements you know—model use, data type, languages or markets, volume, format, timeline, and governance needs.
Scope
We determine the practical project route and the next questions needed to move your dataset work forward.
Frequently asked questions
Using the Data types hub.
Does every data type have a ready-made catalogue?
No. Most BSG DataWorks projects are commissioned. The published pages show the areas you can explore now; a commission brief is the right route when your data needs to reflect a specific language, market, environment, contributor profile, model task, or delivery requirement.
What should I do if I need a data type that is not listed?
Start a dataset brief with the closest data type and explain the intended model use. If the work is within our collection, annotation, or advisory scope, we will help identify the practical next questions.
Can you work with data we already have?
Yes. The annotation and labelling route is for teams that already have source material and need a clear schema, guidelines, quality checks, and delivery structure for a specific AI use case.
How do collection and compliance relate to a data type?
The modality tells you what kind of data a model needs. Collection and compliance support help structure how the project is specified, documented, and reviewed. The appropriate scope depends on the actual project facts.
Can I use this page to request pricing?
Yes. Tell us the data type, model use, approximate volume, timing, and procurement or licensing context if known. If you are still scoping, send what you have and we can clarify the remaining questions on a call.
Start with the modality
Start a dataset brief by email.
Send the outline you have. We will help direct you to the right published page or commissioned project route.
A useful first email includes
- Data type and intended model or product use
- Languages, markets, scenes, domains, or participant profile
- Whether you need new collection, existing data, annotation, or a combination
- Target volume, required format, and any quality requirements
- Rights, governance, consent, or compliance requirements
- Desired timing and approximate procurement or licensing context, if known
If you do not have all of this yet, send what you have. We will fill in the rest on a call.
Prefer to write directly? dataservices@bsgdataworks.com
WhatsApp only: +234 806 790 4903