Speech and voice
Human-sourced speech, voice, transcription, and related language-data work.
About BSG DataWorks
Teaching AI to understand African voices — ethically, and from the ground up.
BSG DataWorks is committed to building ethically sourced, human-centred datasets for AI systems that need to work for African users. We collect, curate, and annotate human-sourced speech, text, image, and video in African-accented English and indigenous African languages—data that has often been under-represented in the material used to train and evaluate AI.
The gap we exist to close
Data that better reflects African voices, languages, and contexts.
Modern AI learns from examples. Yet many widely available datasets give greater weight to American and European varieties of speech and language, leaving African accents, languages, and everyday contexts under-represented. As a result, systems may not reliably reflect how Africans speak and communicate. African languages require deliberate, properly documented collection. That is the gap BSG DataWorks exists to help close.
What we do
Collection, preparation, and licensing support for African AI data.
We provide African-language and African-accented datasets across four data types: speech and voice, text, image, and video. Our work is available through commissioned collection, annotation and labelling, and commercial dataset licensing. Nigerian-accented English and Yorùbá speech samples are available for evaluation; Igbo, Hausa, and Nigerian Pidgin are in the collection pipeline.
Text
Language resources, transcripts, and text data for AI training and evaluation.
Image
Human-generated image data collected and prepared to project requirements.
Video
Real-world video data and client-specific collection projects for African languages and dialects.
We help organisations obtain African data that reflects real languages, markets, and user contexts, supported by documented contributor consent, traceable provenance, and practical quality workflows. Based in Nigeria and working across West Africa, BSG combines linguistic-network experience built since 2005 with focused data collection, preparation, and compliance support for technical and legal teams. Our initial specialization is Nigerian English and Yorùbá, with plans to expand into Nigerian Pidgin, Igbo, Hausa, Kanuri and other African languages.
Who we serve
For teams building technology that needs to understand how Africans communicate.
Our datasets support teams building technology that needs to understand how Africans actually speak and communicate, including the following applications:
Speech recognition and voice AI
Including language-specific transcription, annotation and linguistic QA, voice assistants, call-centre automation for African languages and accents, natural and structured conversations for conversational AI and speech technologies.
Text-to-speech and dubbing
Including African-language voices for media, education, and accessibility.
Language models and translation
Including training and evaluation data for systems serving African users and languages.
Research and academia
Including documented data for African language-technology and NLP research.
AI Training Dataset Collection
Including custom collection of human-generated audio, video, image and language data.
Video Dataset Collection
Client-specific collection projects for African languages and dialects including human activities, lifestyle scenarios and other real-world video data.
Dataset Quality Assurance
Including technical, linguistic, content and metadata QA.
Dataset Licensing
Commercial licensing of BSG DataWorks datasets to AI and technology companies.
What sets us apart
A responsible route to African AI-training data.
Consent-based, always.
Every dataset is sourced from real, consenting contributors—never scraped from the internet or artificially generated.
Compliance-ready.
BSG DataWorks is registered with the Nigeria Data Protection Commission (NDPC/DCP/13596), with data-protection practices designed to support serious buyer review.
Genuine local depth.
We work through native-speaker and linguist networks to source data with the language, accent, and contextual detail a project requires.
Who you are dealing with
A language-services foundation with technical delivery support.
Olayemi Olabenjo
Managing Consultant
- 20+ years language operations
- African-language expertise
- Translation and localisation
- Speech data and annotation
- AI and data-science requirements
- Linguist and contributor networks
BSG DataWorks was founded by Olayemi Olabenjo, Managing Consultant. For more than two decades, he has led Nigerian language-operations work spanning translation, localisation, transcription, speech-data collection, and annotation. This gives BSG a practical foundation for organising the people, language expertise, consent records, and quality checks that an African data-collection project requires.
His experience connects African-language expertise and contributor-network coordination with the AI and data-science requirements behind training and evaluation data. The result is an operating approach that treats collection, annotation, documentation, and quality assurance as connected parts of one delivery process.
The founder is joined by four fellow directors, three of whom are software engineers. Together with a small technical team, they provide the operational and technical support behind collection, quality control, and delivery.
Together, the team is focused on doing this work properly: with genuine consent, careful curation, and compliance built in from the start.
Work with us
Start a dataset brief by email.
Whether you need a commissioned collection, standalone annotation, or a licensed dataset, BSG DataWorks would be glad to help. The best place to begin is a short dataset brief.
Start a dataset brief by email: dataservices@bsgdataworks.com
WhatsApp only: +234 806 790 4903