Kokoro-FastAPI
Dockerized OpenAI-compatible text-to-speech server for Kokoro-82M
What it is
A Docker-packaged FastAPI server around the Kokoro-82M text-to-speech model. It exposes OpenAI-compatible speech endpoints plus websocket interfaces for voice-agent frameworks, with voice mixing, timestamped captions and an optional web UI. Prebuilt images cover CPU, NVIDIA and AMD setups; for self-hosters who need local speech synthesis.
Good for
- Self-host a drop-in OpenAI TTS endpoint
- Generate narration with word timestamps
- Give a local voice agent a TTS backend
Install or run
docker run -p 8880:8880 ghcr.io/remsky/kokoro-fastapi-cpu:latestCommand as given in the project README. Read the repository before you run anything on your machine.
What you need
Docker; an NVIDIA or AMD GPU is optional
Sources
- https://github.com/remsky/Kokoro-FastAPI (project README, read 10 Nov 2026)
- https://api.github.com/repos/remsky/Kokoro-FastAPI (GitHub API, read 13 Nov 2026)
The summary is ours, written from the README. Spotted an error? See how we work.