An open-source Python toolkit for:
- ASR transcription to English from multilingual videos
- English to Simplified Chinese translation
- EN-only / ZH-only / bilingual SRT generation
- Post-cleaning to reduce obvious ASR hallucinations
- Hard-burn bilingual subtitles into video files with ffmpeg
This repository provides a powerful, automated pipeline for generating and hard-burning bilingual subtitles. It was originally developed as part of Ivan's "AI Startup Chronicle" (specifically for the Father Love Global Education video project) but is designed as a modular engine for any video subtitling needs.
This project includes a practical quality rubric inspired by communicative language teaching principles often associated with Prof. Wang Qiang:
- Authenticity: preserve real meaning and context, avoid over-editing speakers.
- Comprehensibility: short, clear lines with synchronized timing.
- Cultural awareness: keep culturally specific references explicit when possible.
- Learning value: bilingual pairing supports noticing and comparison.
- Output quality: remove repetitive hallucinations and check timing consistency.
- Python 3.10+
- ffmpeg installed and available in
PATH(or pass--ffmpeg-bin) - Internet access for translation API calls and model download on first run
python3 -m venv .venv
source .venv/bin/activate
pip install --upgrade pip
pip install -e .Or:
pip install -r requirements.txtautosub burn \
--video-dir ./videos \
--pattern "*.mp4" \
--font "Hiragino Sans GB" \
--skip-existingOutputs:
name.burned.mp4
Follow these steps to push this project to your GitHub:
-
Initialize Git:
git init git add . git commit -m "Initial release of Father Love Subtitle Studio"
-
Create Repo & Push: Go to GitHub, create a new repository called
father-love-subtitle-studio, then run:git branch -M main git remote add origin https://github.com/YOUR_USERNAME/father-love-subtitle-studio.git git push -u origin main
Tip
Make sure NOT to add the large .mp4 video files to Git. They are already in .gitignore. Only scripts and documentation should be uploaded.
- Do not commit copyrighted videos unless you own rights.
- Respect local laws and platform terms when downloading or distributing media.
- Machine transcription and translation require human review for high-stakes use.
MIT - see LICENSE.