A self-hosted, browser-based utility for file conversion, OCR and audio transcription. It wraps common CLI and Python converters (FFmpeg, LibreOffice, Pandoc, ImageMagick, etc.), plus faster-whisper and Tesseract OCR.

Consider spending 30 seconds on the » usage survey « to help improve the app and suggest changes!

• Convert between many file formats; extendable via settings.yml to add any CLI tool.
• OCR for PDFs and images (tesseract / ocrmypdf).
• Audio transcription using Whisper models.
• Simple, responsive dark UI with drag-and-drop and file picker.
• Background job processing with real-time status updates and persistent history.
• /settings page for configuring conversion tools and OAuth (runs without auth in local mode).
• CPU-only by default; a -cuda image is available for GPU use.

Warning: exposing this app publicly without authentication risks arbitrary code execution. Intended for local use or behind a properly configured OAuth/OIDC provider.

FastAPI backend, vanilla HTML/JS/CSS frontend (lightweight), Huey for task queuing, SQLite for storage.

Images available:

• loredcast/filewizard:latest (newest full release without cuda)
• loredcast/filewizard:0.3-small (omits TeX and other large tools)
• loredcast/filewizard:0.3-cuda (CUDA-enabled)

Copy docker-compose.yml from the repo, adjust as needed, then:

docker compose up -d

For different build configurations, use the BUILD_TYPE argument:

# Full build (includes all dependencies but no CUDA)
docker build --build-arg BUILD_TYPE=full -t filewizard:full .

# Small build (excludes TeX and markitdown dependencies for smaller image)
docker build --build-arg BUILD_TYPE=small -t filewizard:small .

# CUDA build (includes CUDA support for GPU acceleration)
docker build --build-arg BUILD_TYPE=cuda -t filewizard:cuda .

Or with docker-compose:

# For full build
docker compose build --build-arg BUILD_TYPE=full

# For small build
docker compose build --build-arg BUILD_TYPE=small

# For CUDA build
docker compose build --build-arg BUILD_TYPE=cuda

For CUDA builds, ensure you have:

• NVIDIA Docker runtime installed (nvidia-docker2 package)
• Compatible GPU with appropriate drivers
• Add the GPU configuration to docker-compose.yml if building with compose:

    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: all
              capabilities: [gpu]

For troubleshooting GPU issues, make sure:

1. Your GPU drivers support the CUDA version (12.1)
2. cuDNN libraries are properly installed in the container
3. The nvidia-container-toolkit is properly configured
4. Test NVIDIA setup with: docker run --rm --gpus all nvidia/cuda:12.1-base-ubuntu22.04 nvidia-smi

git clone https://github.com/LoredCast/filewizard.git
cd filewizard
docker compose up --build

Note: building can be slow (TeX and other dependencies).

git clone https://github.com/LoredCast/filewizard.git
cd filewizard
python -m venv venv
source venv/bin/activate   # Windows: venv\\Scripts\\activate
pip install -r requirements.txt
chmod +x run.sh
./run.sh

Dependencies include fastapi, uvicorn, sqlalchemy, huey, faster-whisper, ocrmypdf, pytesseract, python-multipart, pyyaml, etc.

See the project Wiki for details and examples:
https://github.com/LoredCast/filewizard/wiki

1. Open http://127.0.0.1:8000.
2. Drag & drop or choose files.
3. Select action: Convert, OCR, or Transcribe.
4. Track job progress in the History table (updates automatically).