Open Dubbing

What Open Dubbing is

Open Dubbing is a free, open-source command-line project that uses machine learning to translate spoken dialogue and create a synchronized dubbed video. It runs locally and follows a sequence of separating speech from background audio, identifying speakers, transcribing and translating dialogue, synthesizing voices, and assembling the video. Users can edit generated utterance details, including translated text, voice gender, and speed, then rebuild the video. The tool can also add original-language or dubbed subtitles as streams in the output. Speech recognition options include faster-whisper and Whisper implementations from Hugging Face Transformers; translation can use Meta NLLB or an Apertium-compatible API. Text-to-speech choices include MMS, Coqui, OpenAI, Microsoft Edge, configured command-line tools, and compatible APIs. Available languages depend on the chosen speech and translation systems. FFmpeg is required, and Coqui TTS additionally needs espeak-ng. The project warns that automatic stages can introduce errors and lists support for multiple video input formats as roadmap work. Softcatalà also offers a free test service for English or Spanish videos dubbed into Catalan, with MP4 uploads up to 2 GB or 60 minutes.