Haitian Creole speech dataset
Haitian Creole speech from consented speakers in Haiti and the diaspora, transcribed in the IPN standard orthography, with mobile-band variants suited to the low-bandwidth channels the language is actually used on.
| Language | Haitian Creole |
|---|---|
| Endonym | Kreyòl Ayisyen |
| ISO 639-3 | hat |
| Speaker population | ~12 million |
| Primary regions | Haiti; diaspora in the US, Dominican Republic, Canada, France |
| Dialects covered | Northern, Central (Port-au-Prince), Southern |
| Hours available / collectable | 400 hours available · up to 1,200 hours/quarter collectable |
| Recording spec | 48 kHz / 24-bit WAV plus mobile-band captures, mono |
| Transcript availability | IPN standard orthography, French loanword normalisation notes |
| Consent basis | Recorded verbal consent in Kreyòl, plus written form where literacy allows |
| License terms | Perpetual, worldwide, non-exclusive; humanitarian-use carve-outs available |
| Related languages | Hausa, Zulu, Latin American Spanish, Gulf Arabic, Tagalog |
Why Haitian Creole data is hard to find
Most text labelled Haitian Creole online is machine-translated French, which poisons scraped corpora. Orthography varies widely between the IPN standard and French-influenced spellings, so transcripts need explicit normalisation. Reliable collection depends on field teams with working power and connectivity.
What we can collect
- —Read and spontaneous speech across northern, central and southern varieties
- —Mobile and radio-band audio for real-channel robustness
- —Health, disaster-response and civic-service domain speech
- —Kreyòl–French and Kreyòl–English translation pairs
Licensing and consent
Recorded verbal consent in Kreyòl, plus written form where literacy allows. Delivery is under perpetual, worldwide, non-exclusive; humanitarian-use carve-outs 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 Haitian Creole data at scale?
Send us hours, dialect mix and domain — we'll come back with a collection plan.
Talk to our sourcing team