Orpheus TTS (Pinokio)

Standalone Text-to-Speech using Orpheus TTS (GGUF via llama-cpp-python), SNAC decoding, and a Gradio web UI. Launcher scripts live in the repo root; application code is under app/.

What it does

• Downloads and runs Orpheus multi-language GGUF models with a local Gradio interface.
• Uses a Pinokio-managed Python virtual environment (app/env/).

Using in Pinokio

1. Install — installs dependencies (including PyTorch via torch.js and llama-cpp-python, with CUDA build when an NVIDIA GPU is detected).
2. Start — launches app/app.py on the next free port ({{port}}) and opens the local URL when Gradio prints it.
3. Update — git pull and refreshes Python packages from app/requirements.txt.
4. Reset — removes the app/env/ and app/models/ folders so you can reinstall cleanly.

Programmatic access

The Gradio app listens on 127.0.0.1 at the port shown in the Pinokio “Open Web UI” link (or in the start script logs). Gradio exposes HTTP routes for its UI and API; see the Gradio API docs for building clients.

Example: open the app base URL in a browser after Start completes.

# Replace PORT with the port from the launcher / logs
curl -sS "http://127.0.0.1:PORT/"

For scripted use from Python or JavaScript, use the official Gradio client libraries against the same base URL and port.

Project layout

project-root/
├── app/
│   ├── app.py
│   └── requirements.txt
├── install.js, start.js, update.js, reset.js, link.js, torch.js
├── pinokio.js, pinokio.json
└── README.md