dubb-cli

Local video dubbing with a web UI. Uses faster-whisper (CTranslate2) for transcription and supertonic for TTS — all on-device, no cloud.

CLI

dubb-cli dub <video> --to <lang> -o output.mp4
dubb-cli serve --port 8080
dubb-cli version

Web UI

dubb-cli serve --port 8080
# Open http://localhost:8080/ui

Upload a video, pick a language, download the result.

Pipeline

1. Extract audio (ffmpeg)
2. Transcribe (faster-whisper, CPU int8 with CTranslate2)
3. Translate (Google Translate via deep-translator)
4. Generate speech (supertonic TTS — 31 languages, 10 voices)
5. Speed up video per-segment to match TTS pace
6. Concatenate clips

Requirements

• Python 3.10+
• ffmpeg
• 8 GB+ RAM recommended

Live Demo

https://dubb.dk1.intrane.fr/ui

Performance

Tested on an Intel i5-12400T (10 cores, 8 GB RAM VM). A 40 MB, 5-minute video completed in ~120 seconds (Whisper base model, English → English re-voice).