Hausa speech dataset
Transcribed Hausa speech across the major northern Nigerian dialects, available now and expandable on demand — read, conversational and telephony audio with Boko-orthography transcripts and optional tone marking.
| Language | Hausa |
|---|---|
| Endonym | Harshen Hausa |
| ISO 639-3 | hau |
| Speaker population | ~80 million (L1 + L2) |
| Primary regions | Northern Nigeria, Niger; Ghana, Cameroon, Chad, Sudan |
| Dialects covered | Kano (standard), Sokoto, Katsina, Zaria, Bauchi, Ghanaian Hausa |
| Hours available / collectable | 800 hours available · up to 2,500 hours/quarter collectable |
| Recording spec | 48 kHz / 24-bit WAV plus 8 kHz telephony variants, mono |
| Transcript availability | Boko (Latin) orthography with tone and vowel-length marking on request |
| Consent basis | Written or recorded verbal consent, translated into Hausa before signing |
| License terms | Perpetual, worldwide, non-exclusive; per-domain exclusivity available |
| Related languages | Zulu, Swahili, Yoruba, Burmese, Gulf Arabic |
Why Hausa data is hard to find
Hausa is tonal and has vowel-length contrasts that standard Latin orthography does not mark, so most public transcripts are ambiguous for training. Existing corpora skew heavily toward Kano-dialect radio, missing the Sokoto and Ghanaian varieties. Recruiting rural speakers requires local field teams rather than crowdsourcing platforms.
What we can collect
- —Dialect-balanced read speech across Kano, Sokoto, Katsina and Zaria
- —Telephony-band call audio for ASR in low-bandwidth deployments
- —Agricultural and health-domain spontaneous speech
- —Hausa–English translation and code-switched utterance pairs
Licensing and consent
Written or recorded verbal consent, translated into Hausa before signing. Delivery is under perpetual, worldwide, non-exclusive; per-domain exclusivity available. Call and conversational audio is redacted for personal information before hand-off, and every recording carries a documented consent chain you can audit.
Need Hausa data at scale?
Send us hours, dialect mix and domain — we'll come back with a collection plan.
Talk to our sourcing team