dice-demo.mp4

Speech Cabinet

Create vertical videos in the style of Disco Elysium dialogues.
speech-cabinet.com

Speech Cabinet is an unofficial fan project inspired by Disco Elysium. I am not affiliated with ZA/UM. Disco Elysium© and all related characters, artwork, and audio are property of ZA/UM. These assets are used here for non-commercial, fan purposes only. The code for this app is open-source (AGPL-3.0), but Disco Elysium assets remain under their original copyright. Generated videos are likewise fan-made and not for commercial use.

Tips

• Put quotes when characters speak out loud, like the game does.
• Choose an OST for the right vibes.
• You won’t lose your dialogue if you close the page, but you can always download the file to share or open it later.
• Add custom characters and skills; choose portraits for Harry and custom characters.
• You can self-host via Docker; see Development > Using Docker.
• You should build Communism -- precisely *because* it's impossible.

Development

How it works

This is a Next.js app.

Animation is made purely with CSS/JS, but the videos are rendered on server: a worker starts a browser instance and records a webpage.

Running

1. Create a Postgres database.

2. Copy .env.example to .env and fill in the variables. (Auth settings are not used for now.)

3. If you want to enable background music for videos, save OST music files to public/music. They must have a name in form Sea Power - Instrument of Surrender.m4a.

4. Install dependencies:

   yarn

5. Run:

   yarn dev

6. Run the video rendering worker separately:

   yarn dev:work

   (The worker requires maximum Node 20 because the library for rendering videos uses a deprecated function.)

Using Docker

Requires Docker Desktop (or any Docker Engine + Compose v2). Tested on Apple Silicon via linux/amd64 emulation and on linux/amd64 hosts.

1. cp .env.example .env and fill in NEXTAUTH_SECRET (e.g. openssl rand -base64 32). DATABASE_URL is already wired to the compose db service.
2. (Optional) drop .m4a files into public/music/ if you want music in the rendered video. They must be named exactly as listed in src/lib/music.ts. Missing files are greyed out in the Music menu.
3. Build and start web + worker + postgres:

   DOCKER_DEFAULT_PLATFORM=linux/amd64 docker compose up -d --build

4. Open http://localhost:3000.
5. To tear down: docker compose down, or down -v to also wipe the DB and rendered videos.

Troubleshooting

• macOS Unable to locate package google-chrome-stable: the Chrome apt repo has no arm64 binary. Use DOCKER_DEFAULT_PLATFORM=linux/amd64 as shown above to build for amd64 under Rosetta / QEMU.
• First web request is slow: entrypoint.sh runs yarn db:push on first boot to create the schema.
• Render gets stuck at "Rendering..." on a fresh install: that was issue #20. The worker tried to fetch a gitignored music track and silently 404’d. From this version on, the default is No music and the Music menu disables tracks whose files are not installed. If a render still finishes without music (e.g. the file was removed between selection and render), the UI tells you so.

Contributions

Suggestions and contributions are welcome!