# Dia

> Dia is an open-source text-to-speech model by Nari Labs that generates realistic dialogue audio with multiple speakers, emotions, and non-verbal sounds from transcripts.

Dia is an open-source 1.6B parameter text-to-speech model developed by Nari Labs, designed to generate highly realistic dialogue directly from transcripts. It supports multi-speaker audio generation, non-verbal cues like laughter and coughing, and fine-grained emotion and tone control. Dia can also perform voice cloning using an audio reference, making it a powerful tool for content creators, researchers, and developers building conversational AI applications.

- **Multi-speaker dialogue generation**: *Generate realistic conversations between multiple speakers directly from a text transcript using speaker tags.*
- **Non-verbal audio support**: *Include sounds like laughter, coughing, and sighs in generated audio by adding special tokens in the transcript.*
- **Emotion and tone control**: *Guide the emotional delivery of speech through natural language descriptions embedded in the transcript.*
- **Voice cloning**: *Provide an audio reference clip to clone a specific voice and use it in generated dialogue.*
- **Open-source model weights**: *Download and run the 1.6B parameter model locally via Hugging Face or the GitHub repository.*
- **Gradio demo**: *Try Dia instantly through the hosted Hugging Face Spaces demo without any local setup.*
- **Python API**: *Integrate Dia into your own applications using the provided Python package and inference scripts.*
- **Local inference**: *Run the model on your own hardware for full control over privacy and customization.*

## Features
- Multi-speaker dialogue generation
- Non-verbal audio cues (laughter, coughing, sighs)
- Emotion and tone control via transcript
- Voice cloning from audio reference
- 1.6B parameter open-source model
- Hugging Face Spaces demo
- Python API
- Local inference support

## Integrations
Hugging Face, Gradio

## Platforms
WINDOWS, MACOS, LINUX, WEB, API, DEVELOPER_SDK

## Pricing
Open Source

## Links
- Website: https://github.com/nari-labs/dia
- Documentation: https://github.com/nari-labs/dia/blob/main/README.md
- Repository: https://github.com/nari-labs/dia
- EveryDev.ai: https://www.everydev.ai/tools/dia-tts