Three separate things can be licensed here — a growing dataset, a trained voice model, and the platform that built both. Whatever your use case, get in touch.
Consented voice recordings and transcripts (Roman, Devanagari, and English) from real Fiji Baat speakers. Useful for research, linguistics, or building your own language technology.
The trained Fiji Baat text-to-speech voice itself, for embedding in an app, broadcast, IVR system, or dubbing workflow.
The crowdsourcing and training system behind Fiji Baat works for any dialect or minority language — available to license for other communities and organisations.
A few examples of how these have come up in conversation so far.
Tell us what you're looking for — we'll follow up directly.
