liuzhao1225/YouDub-webui

What it solves

YouDub WebUI is an open-source tool designed to automate the localization of videos from platforms like YouTube and Bilibili. It eliminates the manual effort of separating audio, translating subtitles, and recording new voiceovers, allowing creators to quickly convert videos into different languages while preserving the original background music and sound effects.

How it works

The project implements a complete end-to-end pipeline:

1. Download: Uses yt-dlp to fetch videos from URLs.
2. Audio Separation: Employs Demucs to isolate vocals from background noise.
3. Transcription: Uses openai-whisper to identify speech and generate word-level timestamps.
4. Translation: Translates text via an OpenAI-compatible Chat Completions API.
5. Voice Generation: Uses VoxCPM2 to generate target language dubbing based on reference audio.
6. Final Assembly: Aligns the new voiceover duration, mixes it with the original background audio, and uses FFmpeg to burn in subtitles and export the final MP4.

Who it’s for

It is built for individual content creators, developers, and small teams who want a local, controllable video localization pipeline for cross-language content distribution.

Highlights

• End-to-End Automation: Handles everything from the initial URL to the final rendered video.
• Local-First Approach: Stores all intermediate products, cookies, and logs locally via SQLite.
• Failure Recovery: Supports resuming failed tasks from the specific stage where they stopped.
• Crossover Support: Specifically optimized for English-to-Chinese (YouTube) and Chinese-to-English (Bilibili) workflows.
• Customizable Pipeline: Modular architecture allows users to easily replace ASR, TTS, or translation backends.