Quickstart

Quickstart#

Prerequisites#

  • Python 3.12+

  • A CUDA-capable GPU for most models (CPU-only runs are not guaranteed)

Install#

From PyPI:

pip install vox-serve

From source:

git clone https://github.com/vox-serve/vox-serve.git
cd vox-serve
pip install -e .

Run the server#

vox-serve --model <model-name> --port 8000

Or with the module entrypoint:

python -m vox_serve.launch --model <model-name> --port 8000

Send a request#

Text-to-speech (streaming):

curl -X POST "http://localhost:8000/generate" \
  -F "text=Hello world" \
  -F "streaming=true" \
  -o output.wav

Speech-to-speech (when the model supports audio input):

curl -X POST "http://localhost:8000/generate" \
  -F "text=Hello world" \
  -F "audio=@input.wav" \
  -F "streaming=true" \
  -o output.wav

Health check:

curl -X GET "http://localhost:8000/health"

Notes#

  • streaming=true returns a streaming WAV response with audio chunks.

  • streaming=false returns a single WAV file after the request completes.

  • When using data-parallel or disaggregation modes, ensure you have enough GPUs.

Contents