Data as Taonga: Māori Text-to-Speech Model Reclaims Indigenous Voice from Big Tech

“Our data is a treasure to be guarded, not a commodity to be mined.” This is the philosophy driving Te Hiku Media, a Māori organization that has developed a high-quality text-to-speech (TTS) model tailored specifically to their language and culture. Unlike the generalized models produced by Silicon Valley, this indigenous-owned AI prioritizes linguistic nuances and cultural safety, ensuring the Māori voice is not diluted or exploited.

This initiative marks a significant step in the movement for Indigenous Data Sovereignty. By refusing to hand over their linguistic heritage to Big Tech’s vast datasets, the Papa Reo project demonstrates how marginalized communities can build sophisticated technology on their own terms. The success of this model provides a powerful template for other indigenous groups worldwide looking to harness AI without sacrificing their cultural identity.