kokoro-tts-fastapi

kokoro-tts-fastapi

CONTAINERS IMAGES RUN BUILD

KokoroTTS FastAPI

This Dockerfile sets up a text-to-speech (TTS) service based on the Kokoro TTS engine using FastAPI (https://github.com/remsky/Kokoro-FastAPI)

The server will automatically be started at 0.0.0.0:8880 when you run the container:

docker run -it --rm -p 8880:8880 dustynv/kokoro-tts:fastapi-r36.4.0

Included is a built-in web UI (http://0.0.0.0:8880/web) that converts text to speech using the Kokoro TTS model with GPU acceleration.

Warning

There appears to be an issue with audio playback in the web UI on Firefox - Chrome or Chromium is recommended.

This is the default CMD executed at container startup:

uvicorn api.src.main:app --reload --host 0.0.0.0 --port 8880

CONTAINERS

kokoro-tts:fastapi
Requires	L4T ['>=32.6']
Dependencies	build-essential pip_cache:cu126 cuda:12.6 cudnn python numpy cmake onnx pytorch:2.8 torchvision huggingface_hub rust transformers torchaudio kokoro-tts:hf ffmpeg
Dockerfile	Dockerfile
Images	dustynv/kokoro-tts:fastapi-r36.4.0 (2025-02-19, 5.4GB) dustynv/kokoro-tts:fastapi-r36.4.0-cu128-24.04 (2025-03-03, 4.8GB)

RUN CONTAINER

To start the container, you can use jetson-containers run and autotag, or manually put together a docker run command:

# automatically pull or build a compatible container image
jetson-containers run $(autotag kokoro-tts-fastapi)

# or if using 'docker run' (specify image and mounts/ect)
sudo docker run --runtime nvidia -it --rm --network=host kokoro-tts-fastapi:36.4.0

^{jetson-containers run forwards arguments to docker run with some defaults added (like --runtime nvidia, mounts a /data cache, and detects devices)
autotag finds a container image that's compatible with your version of JetPack/L4T - either locally, pulled from a registry, or by building it.}

To mount your own directories into the container, use the -v or --volume flags:

jetson-containers run -v /path/on/host:/path/in/container $(autotag kokoro-tts-fastapi)

To launch the container running a command, as opposed to an interactive shell:

jetson-containers run $(autotag kokoro-tts-fastapi) my_app --abc xyz

You can pass any options to it that you would to docker run, and it'll print out the full command that it constructs before executing it.

BUILD CONTAINER

If you use autotag as shown above, it'll ask to build the container for you if needed. To manually build it, first do the system setup, then run:

jetson-containers build kokoro-tts-fastapi

The dependencies from above will be built into the container, and it'll be tested during. Run it with --help for build options.