ReFlow Studio is a local, privacy-focused AI video production pipeline. It automates the complex tasks of dubbing, censoring, and editing videos using state-of-the-art machine learning models, all wrapped in a premium "Glass/Dark" dashboard.
Note: This tool processes all data locally on your GPU. No data is sent to the cloud.
- Engine: Uses Coqui XTTS v2 for high-fidelity voice cloning and dubbing.
- Workflow: Transcribes original audio, translates text, and generates dubbed audio in the target language (Hindi, English, etc.) while preserving the original voice tone.
- Documentary Mode: Optional "Duck-and-Cover" mixing (keeps original background audio at 20% volume).
- Audio: Uses OpenAI Whisper to transcribe text and
better_profanityto identify offensive keywords. Automatically mutes or "bleeps" specific timestamps. - Visual: Integrates NudeNet to detect NSFW visual content frame-by-frame and applies a Gaussian blur to specific regions automatically.
- Built with CustomTkinter for a high-DPI, responsive interface.
- Queue System: Batch process multiple videos in one go.
- Live Stats: Real-time tracking of processing speed, total jobs, and system status.
- Dual Theme: One-click toggle between "Deep Night" (Cyberpunk) and "Clean Light" modes.
- GUI: Python CustomTkinter (Modern UI)
- Audio AI: OpenAI Whisper (Transcription), Coqui XTTS v2 (Dubbing)
- Vision AI: NudeNet (Object Detection)
- Processing: FFmpeg (Media merging and encoding)
- Backend: PyTorch (Tensor computation)
- Windows 10/11 (NVIDIA GPU recommended for speed)
- Python 3.10+
- FFmpeg installed and added to System "PATH".
-
Clone the repository
git clone [https://github.com/ananta-sj/ReFlow-Studio.git](https://github.com/ananta-sj/ReFlow-Studio.git) cd ReFlow -
Install Dependencies (Note: Some AI models require specific PyTorch versions)
pip install -r requirements.txt
-
Run the App
python main.py
ReFlow Studio is a technical demonstration of local AI capabilities.
- Model Licenses: This project uses Coqui XTTS (CPML) and NudeNet. Ensure you comply with their specific licenses if you intend to use this software for commercial purposes.
- Content Usage: The developer is not responsible for any misuse of this software to alter content without consent. Please use responsibly.
| Dark Mode | Light Mode |
|---|---|
 |
 |
1. "FFmpeg not found" Error:
- Make sure you downloaded FFmpeg and added the
binfolder to your Windows System PATH. - Restart your computer after installing FFmpeg.
2. App crashes on startup:
- Check if you have the latest Visual C++ Redistributables installed.
- Ensure your NVIDIA drivers are up to date (if using GPU).
3. "Visual Blur" is slow:
- NudeNet requires downloading models on the first run. Please be patient; it will be faster next time.
Contributions are welcome!
- Fork the Project
- Create your Feature Branch (
git checkout -b feature/AmazingFeature) - Commit your Changes (
git commit -m 'Add some AmazingFeature') - Push to the Branch (
git push origin feature/AmazingFeature) - Open a Pull Request
Distributed under the MIT License. See LICENSE for more information.