Nari Labs

AI Native

Nari Labs' Dia is an open-weights text-to-speech model family that generates multi-speaker, dialogue-style audio from transcripts with speaker tags, supports expressive non-verbal cues, and is optimized for local GPU-first deployment.