Open Dubbing
Experimental command-line tool that transcribes, translates and re-voices a video's dialogue into another language.
Open Dubbing chains speech recognition, translation and text-to-speech to dub a video into another language.
Open Dubbing is a command-line tool from Softcatala that automatically translates and synchronises a video's spoken dialogue into another language. Its own README calls it pure experimental. It suits developers and localisation hobbyists who want an open pipeline they can inspect and hand-correct.
What it is
You give it a video file and a target language, and it separates vocals, identifies speakers and gender, transcribes with Whisper, translates, synthesises speech and rebuilds the video. Intermediate files include a JSON of utterances that you can edit and re-run with --update to change translated text, voice gender or speed.
Who it's for
It suits people experimenting with automated dubbing, especially for languages such as Catalan, who are comfortable with Python, ffmpeg and Hugging Face tokens. It does not suit anyone needing broadcast-quality dubs or lip-sync, because each stage can introduce errors and the project is marked experimental.
Key features
End-to-end dubbing
Takes a video and produces a dubbed version in the target language.
Selectable TTS engines
Supports Coqui, MMS, Edge and OpenAI TTS, or any other engine through an API or CLI.
Speaker and gender detection
Assigns synthetic voices by detected gender and speaker.
Multiple translation engines
Use Meta's NLLB, the Apertium API and others.
Automatic source language
Whisper detects the source language from the first 30 seconds.
Editable utterance JSON
Fix translations, voice or timing in a JSON file and rebuild with --update.
Up and running in 3 steps
Our honest verdict
What's great
- Post-editing is built in: change translated_text, gender or speed in the utterance JSON and regenerate only the dubbing.
- You can mix components, such as NLLB or Apertium for translation and Coqui, MMS, Edge or OpenAI for speech.
- It is built on named open components (Demucs, pyannote, faster-whisper, NLLB-200), so you can see what runs each stage.
- A live Catalan demo shows what the combined system can do for English and Spanish sources.
Where it falls short
- The README calls it experimental and warns that errors can be introduced at each step: speech recognition, translation and vocal recognition.
- Setup needs pip extras (coqui, openai), ffmpeg, espeak-ng for Coqui, and accepting pyannote model terms on Hugging Face with an access token.
- The roadmap lists optimising for long videos and lower resource use, which implies heavy processing for long inputs; Windows is described as working but not yet widely used.
- Language support depends on the TTS and translation combination you choose, and the author says he only tried very few of the languages listed.
Pricing & license
Open Dubbing is free under Apache-2.0 and the project shows no paid plan. The only cost is your own compute, plus API charges if you choose OpenAI TTS or another paid engine.
Open remote AI-training roles
870 open remote AI-training roles, updated November 14, 2026; most pay $20–$170 an hour. Zekai lists them free; the application goes to the employer’s platform.
Alternatives
These reviewed AI tools do the same job instantly — from our scored directory.
Use it, or have us set it up
FAQ
Can it dub into any language?
The target list depends on your TTS and translation choice. The README lists many languages but says the author tried only very few of them.
Can I correct mistakes in the output?
Yes. Edit utterance_metadata_XXX.json, changing translated_text, gender or speed, then rerun with --update to rebuild the dub.
Is there a web interface?
The README points to a demo at softcatala.org/doblatge and to separate subdub-editor and dubbing-service repositories; this repository itself is a command-line tool.
