Subtitle Edit for Unraid & Docker

Run the native Linux version of Subtitle Edit in a web browser.

Subtitle Edit for Unraid packages the native Linux edition of Subtitle Edit with FFmpeg/FFprobe, MPV/libmpv, Tesseract OCR, optional hardware-accelerated video playback, and the jlesage browser GUI.

The primary tested platform is AMD64 Unraid, but the image can also run on an AMD64 Linux Docker host.

This is an unofficial community project. It is not maintained or endorsed by the Subtitle Edit, Unraid, or jlesage projects.

Contents

• Features
• Quick start on Unraid
• Quick start with Docker
• Docker Compose
• Image tags
• Included software
• CPU and GPU setup
• Browser interface
• Files and persistence
• OCR engines
• Updating
• Network access and security
• Support and development

Features

• Native Linux Subtitle Edit available through any modern browser
• FFmpeg/FFprobe, MPV/libmpv, and Tesseract OCR included
• Browser audio and GTK file chooser integration
• Persistent settings, OCR data, downloaded models, and media mappings
• CPU operation with optional Intel and AMD GPU acceleration
• Unraid template and standard Docker deployment

Quick start on Unraid

The template is being prepared for Community Applications. Until it is listed, install the user template from an Unraid terminal:

curl -fsSL \
  https://raw.githubusercontent.com/pegasbur/subtitleedit/main/unraid/my-subtitleedit.xml \
  -o /boot/config/plugins/dockerMan/templates-user/my-subtitleedit.xml

Then open Docker > Add Container and select Subtitle Edit from the template list.

During installation:

1. Confirm the appdata and media paths.
2. Configure GPU access if required, or leave it disabled for CPU/software operation.
3. Apply the template and open its WebUI.

The default address is:

http://UNRAID-IP:3001

The host port can be changed if 3001 is already used by another container.

Quick start with Docker

Create persistent folders:

mkdir -p /srv/subtitleedit/config /srv/media

Run the container:

docker run -d \
  --name subtitleedit \
  --restart unless-stopped \
  --shm-size=1g \
  -p 3001:5800 \
  -e USER_ID=1000 \
  -e GROUP_ID=1000 \
  -e TZ=Etc/UTC \
  -e WEB_AUDIO=1 \
  -e KEEP_APP_RUNNING=1 \
  -e DISPLAY_WIDTH=1920 \
  -e DISPLAY_HEIGHT=1080 \
  -v /srv/subtitleedit/config:/config \
  -v /srv/media:/data \
  ghcr.io/pegasbur/subtitleedit:latest

Change the host paths, timezone, and numeric user and group IDs for your system. Then open:

http://DOCKER-HOST-IP:3001

For Intel or AMD GPU access, add:

--device=/dev/dri

Docker Compose

Copy the example environment file:

cp .env.example .env

Review the image, container name, WebUI port, appdata path, and media path in .env.

Start in CPU/software mode:

docker compose up -d

Start with Intel or AMD GPU access:

docker compose \
  -f compose.yaml \
  -f compose.gpu.yaml \
  up -d

Image tags

Included software

Exact source commits, package checksums, builder images, and runtime images are pinned in versions.env.

CPU and GPU setup

A GPU is optional. Subtitle Edit remains usable through CPU/software operation without one.

NVIDIA GPU acceleration is not supported by the jlesage GUI stack, so NVIDIA systems use CPU/software fallback.

GPU access can accelerate desktop rendering and supported video decoding. It does not make Tesseract or CPU-based optional OCR engines use the GPU.

Browser interface

The browser interface is provided by the jlesage noVNC-based GUI stack.

• Web access uses container port 5800.
• Native VNC is available on container port 5900, but is not mapped by default.
• The side panel provides audio, clipboard, scaling, quality, and logging controls.
• Remote resizing adapts the virtual desktop to the browser window.

Files and persistence

The principal container paths are the same on Unraid and Docker:

Correctly mapped /config and /data directories remain available when the container is updated or replaced.

Additional host directories can be mapped when required.

OCR engines

The Tesseract OCR engine, English language data, and orientation/script detection data are included.

Additional Tesseract .traineddata files can be placed in:

/config/tessdata

Subtitle Edit stores application-managed Tesseract models in its own configuration directory:

/config/xdg/config/Subtitle Edit/Tesseract550/tessdata

This directory is part of the persistent /config mapping and does not require an additional volume mapping.

Large optional engines and models such as PaddleOCR, CrispEmbed, Whisper, and llama.cpp can be downloaded through Subtitle Edit when requested. Their files are retained under /config.

Tesseract normally uses the CPU. Optional engines use a GPU only when the selected engine and downloaded package support it.

Updating

Unraid

Use Unraid’s normal Check for Updates and Update controls. Settings, downloaded OCR models, bookmarks, and media remain in their mapped directories.

Docker

Pull the current image and recreate the container with the same options and volume mappings:

docker pull ghcr.io/pegasbur/subtitleedit:latest

Docker Compose users can run:

docker compose pull
docker compose up -d

Network access and security

The WebUI is intended for a trusted LAN, Tailscale, or a secured reverse proxy and should not be exposed directly to the public Internet. It uses HTTP without built-in authentication by default (SECURE_CONNECTION=0, WEB_AUTHENTICATION=0), although HTTPS and password authentication can be enabled through the corresponding container settings.

Support and development

Report container-image or Unraid-template problems in this repository’s issue tracker. Application bugs should be reported to the Subtitle Edit project after confirming that they are not specific to this container.

Build instructions, version maintenance, validation, and publishing information are available in docs/DEVELOPMENT.md.

The integration files are MIT licensed. Included software retains its own licenses; see THIRD_PARTY_NOTICES.md.