Kokoro-FastAPI-GPU
ghcr.io/remsky/kokoro-fastapi-gpu:latest
https://github.com/remsky/Kokoro-FastAPI/pkgs/container/kokoro-fastapi-gpu
bridge
bash
false
https://github.com/remsky/Kokoro-FastAPI/discussions/230
https://github.com/remsky/Kokoro-FastAPI
Kokoro-FastAPI is a Dockerized FastAPI wrapper for the Kokoro-82M text-to-speech model, optimized for speed and quality.[br][br]
[b]Features:[/b][br]
- High-quality Japanese and English text-to-speech.[br]
- Accelerated inference using NVIDIA GPUs.[br]
- OpenAI-compatible speech endpoint.[br]
[br]
[b]Home Assistant Integration:[/b][br]
This container is fully compatible with the [b]OpenAI TTS[/b] integration in Home Assistant. You can generate local, OpenAI-level quality voice audio using your own GPU.[br]
1. Install the OpenAI TTS integration (via HACS).[br]
2. Set the "API Key" to any random text (e.g., "123").[br]
3. Set the "Target URL" to: [b]http://[IP]:8880/v1[/b][br]
[br]
[b]Requirements:[/b][br]
- Requires the [b]Nvidia Driver[/b] plugin.[br]
- This container runs as internal user [b]1001[/b]. Ensure your "Temp Files" output directory is writable by this user.[br]
- The WebUI is located at [b]/web[/b].
AI: Tools:Utilities
http://[IP]:[PORT:8880]/web
https://github.com/ds-sebastian/unraid-templates/blob/main/icons/kokoro-icon.png?raw=true
--gpus all --restart=unless-stopped
Nvidia Driver
8880
all
all
true
gpu