Skip to main content
Version: v5.0.0

Kokoro-FastAPI

Run Kokoro-FastAPI locally and connect it to OpenReader using the Custom OpenAI-Like provider.

warning

For Kokoro issues and support, use the upstream repository: remsky/Kokoro-FastAPI.

Run Kokoro​

CPU:

docker run --name kokoro-tts \
  --restart unless-stopped \
  -d \
  -p 8880:8880 \
  -e ONNX_NUM_THREADS=8 \
  -e ONNX_INTER_OP_THREADS=4 \
  -e ONNX_EXECUTION_MODE=parallel \
  -e ONNX_OPTIMIZATION_LEVEL=all \
  -e ONNX_MEMORY_PATTERN=true \
  -e ONNX_ARENA_EXTEND_STRATEGY=kNextPowerOfTwo \
  -e API_LOG_LEVEL=DEBUG \
  ghcr.io/remsky/kokoro-fastapi-cpu:v0.2.4

GPU (NVIDIA):

docker run --name kokoro-tts \
  --restart unless-stopped \
  -d \
  --gpus all \
  --user 1001:1001 \
  -p 8880:8880 \
  -e USE_GPU=true \
  -e PYTHONUNBUFFERED=1 \
  -e API_LOG_LEVEL=DEBUG \
  ghcr.io/remsky/kokoro-fastapi-gpu:v0.2.4

Connect to OpenReader​

Recommended (auth + admin): Settings → Admin → Shared providers

  1. Add a shared provider with type custom-openai.
  2. Set the base URL for your deployment topology (Docker-to-host: http://host.docker.internal:8880/v1; native same-host: http://127.0.0.1:8880/v1).
  3. Leave API key blank unless required by your deployment.
  4. Set default model to Kokoro.

Bootstrap seed (optional, first boot only):

API_BASE=http://host.docker.internal:8880/v1
API_MODEL_NAME=kokoro

Use host.docker.internal only when OpenReader/its embedded worker run in Docker and Kokoro runs on that Docker host. In Compose, use the shared service URL http://kokoro-tts:8880/v1. A remote worker needs a public/private-network URL it can reach. See the provider topology table.

Users select the configured shared provider, model, and voice from Settings → TTS Provider.

References​