Offline translation

Starting from version 5.0, Linguist has a built-in offline translator called "Bergamot". To use it, simply choose this translator on the preferences page.

You can also use offline translation with custom translators starting from Linguist version 4.0.

To do this, deploy any service locally for translating text and implement a JS binding for Linguist.

LibreTranslate

You can use LibreTranslate to deploy a machine translation service locally or on your own server.

See the Installation documentation to deploy LibreTranslate locally via Docker.

Quick start instructions to deploy LibreTranslate locally:

Install Docker
Install the NVIDIA Container Toolkit to allow the container to use your GPU
Create a compose.yml file:

services:
  libretranslate:
    container_name: libretranslate
    image: libretranslate/libretranslate:latest-cuda
    ports:
      - "5000:5000"
    restart: unless-stopped
    command: --disable-web-ui
    environment:
     - LT_API_KEYS_DB_PATH=/app/db/api_keys.db
     - LT_API_KEYS=True
     - LT_UPDATE_MODELS=True
     - LT_DEBUG=True
     - PUID=root
    volumes:
     - ~/.libretranslate/db:/app/db
     - ~/.libretranslate/data:/root/.local:rw
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: 1
              capabilities: [gpu]

Start the container with docker compose up -d --build

Once the container is built, LibreTranslate will download the models. This will take some time. You can check http://localhost:5000/ or view the Docker logs with docker compose logs -f. Once the models are loaded, you can use the translator.

Once you have finished deployment (or if you find a trusted server), add the custom translator LibreTranslator in Linguist settings. If you use a remote instance, replace the apiPath value with the actual URL of the LibreTranslate instance.

Done. You can now translate everything locally.

LibreTranslate​

LibreTranslate