Dia TTS is a 1.6B parameter open-source text-to-speech model by Nari Labs. It generates ultra-realistic, emotionally expressive speech, including non-verbal cues like laughter and sighs. Ideal for creators, developers, and researchers.