katarive-docker

This repository provides a basic Docker setup for Katarive, a system for generating and managing narrated audio content.

Prerequisites

• Docker
• Docker Compose
• Make (optional, but recommended)

Getting Started

The easiest way to start Katarive is using the provided Makefile.

1. Start the Containers

By default, this will start the CPU version of VOICEVOX.

make start

If you have an NVIDIA GPU and the NVIDIA Container Toolkit installed, run:

make start TARGET=nvidia

Once the containers are running, you can access the Web UI at: http://localhost:9421

2. Stop the Containers

make stop

Advanced Usage

Available Make Commands

Command	Description
make build	Build the Docker images.
make start	Build and start the containers in detached mode.
make stop	Stop and remove the containers.

You can specify the hardware target for any command using TARGET=cpu (default) or TARGET=nvidia.

Customizing Plugins

Katarive plugins are automatically built during the Docker image creation based on the configuration in plugins.csv.

To add or remove plugins:

1. Edit plugins.csv.
2. Rebuild the images: make build (or make start).

The format of plugins.csv is: repository_url,binary_name (binary name is optional).

Tip

Plugins with the default- prefix in their binary name have the lowest priority. If you want to override a default plugin's functionality, simply add a custom plugin without the default- prefix.

Manual Execution

If you don't have make, you can run the commands manually. Ensure that UID and GID are set to avoid permission issues with mounted volumes:

export UID=$(id -u)
export GID=$(id -g)
docker compose -f docker-compose.cpu.yaml up -d --build

Customization

• Modifying the docker-compose.*.yaml files to change ports or environment variables.
• Updating the Dockerfile for more complex build changes.
• Mounting your own data volumes for persistent storage.