Skip to content

Latest commit

 

History

History
80 lines (47 loc) · 1.96 KB

README.md

File metadata and controls

80 lines (47 loc) · 1.96 KB
Raw

sbv2-api-lite

Ultra instant compatible API for Style-Bert-VITS2 - You'll have it running before your morning coffee is ready! ☕️

🚀 Quick start

  1. Clone this repo

    git clone https://github.com/uezo/sbv2-api-lite
cd sbv2-api-lite

  2. Install dependencies

    pip install -r requirements.txt

# If you use GPU, reinstall PyTorch that matches the CUDA version of your environment.
pip uninstall torch torchvision torchaudio
pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu124   # for CUDA 12.4

  3. Start the API server

    uvicorn run:app --host 0.0.0.0 --port 5000

Now go to http://127.0.0.1:5000/docs . Upload yoru model file and enjoy high quality TTS🥳

📦　Run on RunPod

Set the following command as the Container Start Command in your Pod console, then start the Pod to execute the process. Once complete, stop the Pod in the same order.

  1. Set up the repository in the workspace

    git clone https://github.com/uezo/sbv2-api-lite /workspace/sbv2-api-lite

  2. Create the runtime environment and install dependencies

    bash /workspace/sbv2-api-lite/runpod_install.sh

  3. Set up the API server to launch at startup

    bash /workspace/sbv2-api-lite/runpod_start.sh

    Add the following environment variables with the value true:

    SBV2_API_LITE_USE_GPU
SBV2_API_LITE_VERBOSE

    If you want to use compressed audio, add the variable with the value ffmpeg-static/ffmpeg:

    SBV2_API_LITE_FFMPEG_PATH

  4. Click Connect to HTTP Service [Port 5000]

    Click Connect to HTTP Service [Port 5000] button and head over to /docs to see the OpenAPI interface.

Upload model file with GPU=True. Enjoy high quality TTS🥳

Note: You only need to configure this setup once. After that, just start the Pod, and the API server will automatically launch.