EveryVoice is a text-to-speech system developed by the National Research Council Canada in partnership with Indigenous communities and educational institutions. The system generates realistic pronunciation recordings for three Indigenous languages: Kanyen'kéha (Mohawk), nēhiyawēwin (Plains Cree), and SENĆOŦEN. These recordings are integrated into language learning materials such as dictionaries and verb conjugation references to support language preservation and education.
This system is currently in development and is designed to serve members of the public, particularly Indigenous language learners and educators. The system does not involve the collection of personal information from users. All training data comes from Indigenous language recordings that were obtained with proper licensing agreements and informed consent from Indigenous communities, ensuring respectful and ethical use of cultural and linguistic knowledge.
As an AI system working with culturally significant linguistic content, EveryVoice operates with strong community partnerships and transparency about its training methods. Users can learn how the system works and understand that their language learning experience is supported by technology developed in collaboration with Indigenous peoples and institutions.
