docker.md

Docker Setup

Harmony Speech Engine can be easily set up and run using Docker. This allows you to run the application without worrying about installing dependencies manually.

Prerequisites

• Docker: Make sure Docker is installed on your system. You can download it from Docker's official website.
• Docker Compose: Ensure Docker Compose is installed. It usually comes with Docker Desktop installations.

For GPU support:

• NVIDIA GPU Drivers: Required if you plan to use GPU acceleration.
• NVIDIA Container Toolkit: Install from NVIDIA Container Toolkit documentation.

Docker Compose Files

There are two docker-compose files provided:

• docker-compose.yml: Uses CPU only. Suitable if you don't have an NVIDIA GPU.
• docker-compose.nvidia.yml: Utilizes NVIDIA GPUs for acceleration.

Running with Docker Compose (CPU Only)

1. Navigate to the project directory:

   cd harmony-speech-engine

2. Start the services:

   docker compose up

This will start both the Harmony Speech Engine API and the frontend UI, using the CPU specific compose-file by default.

Running with Docker Compose (NVIDIA GPU Support)

1. Ensure your system meets the GPU prerequisites mentioned above.
2. Navigate to the project directory:

   cd harmony-speech-engine

3. Start the services with GPU support by specifying the docker-compose file created for using nvidia:

   docker compose -f docker-compose.nvidia.yml up

By default, out images provided via docker hub will be downloaded. However you can also build the docker images yourself locally, by removing the hashtag '#' in front of the build instruction inside the compose files.

If you want to start the docker containers as services in the background, you can use the -d flag:

docker compose -f docker-compose.yml up -d

docker compose -f docker-compose.nvidia.yml up -d

Stopping the Services

To stop the running containers:

docker compose down

or for the NVIDIA GPU setup:

docker-compose -f docker-compose.nvidia.yml down

Configuration

• Model Configuration: The models and their settings are defined in config.yml for CPU usage and config.nvidia.yml for GPU usage.
• Volumes: The Docker setup mounts the config.yml or config.nvidia.yml and the cache directory into the container. Ensure these files and directories exist in your project root

Accessing the Application

• API: Once the services are up, the Harmony Speech Engine API is accessible at http://localhost:12080.
• Frontend UI: Access the UI at http://localhost:8080.

Notes

• Ensure that the ports 12080 and 8080 are available on your host machine.
• If you need to customize the configurations, edit the config.yml or config.nvidia.yml files accordingly.