AI Transcription, Diarization & Summarization Tool

A local, privacy-focused web application that uses advanced AI models to transcribe, differentiate speakers (diarization), and summarize audio/video content.

📋 Prerequisites

Before setting up the project, ensure you have the following installed on your Windows system:

1. Python 3.10 or 3.11 (recommended)
   • Download Python
   • Check "Add Python to PATH" during installation.
2. Microsoft Visual C++ Build Tools (Required for some AI libraries)
   • Download Build Tools
   • Install "Desktop development with C++" workload.
3. FFmpeg (Required for audio processing)
   • Download FFmpeg
   • Extract and add the bin folder to your System PATH.
   • Verify by opening CMD and typing: ffmpeg -version
4. MongoDB Community Server
   • Download MongoDB
   • Install as a Service.

🛠️ Installation Guide

1. Clone the Project

   cd "path\to\AI-summerizztion"

2. Create Virtual Environment

   python -m venv venv
   .\venv\Scripts\activate

3. Install Dependencies

   pip install -r requirements.txt

4. Enable GPU Support (Recommended) To make transcription 10x-50x faster (requires NVIDIA GPU):

   # Uninstall CPU torch first
   pip uninstall torch torchvision torchaudio -y
   
   # Install CUDA 11.8 version
   pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118

5. Setup Authentication (Important)

   A. HuggingFace Token (For Diarization)

   1. Get a token from Hugging Face.
   2. Accept user agreements for pyannote/speaker-diarization-3.1 and pyannote/segmentation-3.0.
   3. Create a .env file in the root directory:

      HF_TOKEN=hf_yourtokenhere...

   B. YouTube Cookies (For Downloads) Fixes "Sign in to confirm you’re not a bot" errors.

   1. Install "Get cookies.txt LOCALLY" extension on Chrome/Firefox.
   2. Go to YouTube (logged in).
   3. Export cookies and save the file as cookies.txt in this project folder (d:\VS Code\Python\AI-summerizztion\cookies.txt).

🚀 How to Run

1. Activate Environment (if not active)

   .\venv\Scripts\activate

2. Start the App

   python app.py

3. Open in Browser

   • Go to: http://127.0.0.1:5000

Note: The first run will download models (approx 2GB). Please be patient.

⚠️ Common Issues

• DownloadError: Sign in to confirm...:
  • Missing cookies.txt. See Step 5B above.
• ModuleNotFoundError:
  • Ensure venv is activated.
• torchcodec Warning:
  • You can safely ignore "torchcodec is not installed correctly" warnings if the app starts.