What we’ve learned shipping voice infrastructure for the world’s underserved languages.
Voice infrastructure for the languages every other model treats as an afterthought — production-grade speech recognition and text-to-speech for the world that never got served.
