Dia

Overview

Dia is a 1.6B parameter text-to-speech model from Nari Labs designed to generate remarkably natural, expressive dialogue directly from a transcript. It can render non-verbal cues such as laughter and sighs, and supports speaker tags for multi-voice conversations. The open weights make it a favourite for building lifelike conversational experiences.

What makes Dia special?

• Focus: Expressive dialogue generation.
• Availability: Free & Open-Source.
• Use cases: Dialogue systems, audiobooks, and conversational agents.

Visit official site