Video to Document Converter (Video2PDF)

2. Install the required dependencies:

   pip install -r requirements.txt

   Note: You may need ffmpeg installed on your system for video processing.

3. GPU Support (Recommended): For faster processing and better accuracy (OCR), install PyTorch with CUDA support:

   pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu121

4. Configuration:

   • Duplicate the example configuration file:

     cp .env.example .env

   • Open .env and add your Gemini API Key (get one from Google AI Studio).

   • This is required for advanced transcript cleaning features.

   • This is required for advanced transcript cleaning features.

5. Ollama (Optional - For AI Transcript Cleaning):

   • Install Ollama.
   • Pull the required model (default is gemma3):

     ollama pull gemma3

   • Ensure Ollama is running (ollama serve).

Advanced Setup

YouTube Authentication (Cookies)

If you encounter "Sign in to confirm you're not a bot" errors:

1. Automatic: Log in to YouTube in Chrome, Edge, or Firefox. The tool attempts to extract cookies automatically.
2. Manual:
   • Export your YouTube cookies to Netscape format (using a browser extension like "Get cookies.txt LOCALLY").
   • Save as cookies.txt.
   • Use the CLI: python main.py URL --cookies cookies.txt.

Usage

Web Interface (Recommended)

Start the web application:

python app.py

Open your browser to http://localhost:5000.

Workflow:

1. Start: Enter a YouTube URL or select an existing project.
2. Process: The system downloads the video, extracts frames, and detects duplicates.
   • Options:
     • Skip Video Download: Use existing video file if available.
     • Skip Extraction: Use existing extracted images.
     • Skip Deduplication: Use existing deduplication results.
     • Download Transcript: Download subtitles/transcript from YouTube.
3. Curate: Review the extracted slides in the "Curate Slides" interface.
   • Green: Kept slides.
   • Red: Detected duplicates (discarded).
   • Yellow: Detected blanks (discarded).
   • Click any slide to toggle its status.
4. Edit: Review the synchronized transcript for each slide.
5. Generate: Download your final PDF or DOCX.

Bulk Transcript Extraction

You can extract transcripts from YouTube playlists or local directories (recursively) using the dedicated tool:

# Process a YouTube Playlist
python bulk_transcribe.py "https://www.youtube.com/playlist?list=PL..." -o ./output

# Process a Local Directory (creates transcripts using Whisper)
python bulk_transcribe.py "./my_videos" -o ./output --model base

Features:

• YouTube Playlists: Downloads all available transcripts/captions.
• Local Videos: Automatically transcribes video/audio files (.mp4, .mp3, etc.) using OpenAI Whisper.
• Recursive: Scans subfolders for local files.
• Models: Supports different Whisper model sizes (tiny, base, small, medium, large).

CLI Usage (Advanced)

You can also run the pipeline from the command line:

python main.py <VIDEO_URL_OR_PATH> [OPTIONS]

Common Options:

• --download-transcript: Download subtitles/transcript from YouTube.
• --create-docx: Generate a DOCX file with synchronized transcript.
• --create-combined: Generate a PDF file with synchronized transcript.
• --optimize-images: Resize and compress images to reduce output file size.

Project Structure

This repository is organized as follows:

• app.py: Main Flask web application.
• main.py: CLI entry point.
• scripts/: Utility scripts (e.g., analyze_images.py, convert_md_to_docx.py) and experimental tools.
• docs/: Documentation files including testing guides and optimization reports.
• experiments/: Source code for previous experiments (deduplication, compression).
• archive/: Deprecated code and temporary files.

Output Structure

The tool organizes outputs into a structured folder:

output/
└── Video Title/
    ├── video/                  # Downloaded video file
    ├── images/                 # Extracted screenshots
    │   └── organized_moderate/ # Curated unique/duplicate folders
    ├── transcripts/            # Raw transcript files
    ├── Video Title.pdf         # Final PDF
    ├── Video Title.docx        # Final DOCX
    └── metadata.txt            # Video metadata

For more detailed documentation, please check the docs/ folder.