nari-labs/dia

What it is

Dia is a 1.6B parameter text to speech model created by Nari Labs. It directly generates highly realistic dialogue from a transcript using [S1] and [S2] speaker tags, and can be conditioned on audio for emotion and tone control. It can also produce nonverbal sounds such as laughter and coughing. Pretrained checkpoints and inference code are provided, and it currently supports English only.

Who it's for

• Researchers and developers experimenting with open-weights dialogue text-to-speech
• Developers who want multi-speaker conversational audio from a scripted transcript
• Users who want voice cloning via audio prompts
• Hugging Face Transformers users who want to run Dia through that library

Requirements

Setup

1. Install from a clone (pip)

   Clone the repo, optionally create a virtual environment, and install in editable mode.

   bashCopy

   # Clone this repository
   git clone https://github.com/nari-labs/dia.git
   cd dia
   
   # Optionally
   python -m venv .venv && source .venv/bin/activate
   
   # Install dia
   pip install -e .

2. Install directly from GitHub

   Install without cloning the repository.

   bashCopy

   # Install directly from GitHub
   pip install git+https://github.com/nari-labs/dia.git

3. Run the simple example

   Run the bundled example after installing with pip.

   bashCopy

   python example/simple.py

4. Run with uv

   With uv installed, clone the repo and run the example directly.

   bashCopy

   # Clone this repository
   git clone https://github.com/nari-labs/dia.git
   cd dia
   
   uv run example/simple.py

5. Install Transformers main branch

   Needed to use the Hugging Face Transformers implementation of Dia.

   bashCopy

   pip install git+https://github.com/huggingface/transformers.git
   # or install with uv
   uv pip install git+https://github.com/huggingface/transformers.git

Examples

from transformers import AutoProcessor, DiaForConditionalGeneration


torch_device = "cuda"
model_checkpoint = "nari-labs/Dia-1.6B-0626"

text = [
    "[S1] Dia is an open weights text to dialogue model. [S2] You get full control over scripts and voices. [S1] Wow. Amazing. (laughs) [S2] Try it now on Git hub or Hugging Face."
]
processor = AutoProcessor.from_pretrained(model_checkpoint)
inputs = processor(text=text, padding=True, return_tensors="pt").to(torch_device)

model = DiaForConditionalGeneration.from_pretrained(model_checkpoint).to(torch_device)
outputs = model.generate(
    **inputs, max_new_tokens=3072, guidance_scale=3.0, temperature=1.8, top_p=0.90, top_k=45
)

outputs = processor.batch_decode(outputs)
processor.save_audio(outputs, "example.mp3")

python app.py

# Or if you have uv installed
uv run app.py

python cli.py --help

# Or if you have uv installed
uv run cli.py --help

[S1] Dia is an open weights text to dialogue model. [S2] You get full control over scripts and voices. [S1] Wow. Amazing. (laughs) [S2] Try it now on Git hub or Hugging Face.

Pros & cons