TubeInsight

TubeInsight is a Python-based tool that extracts YouTube video transcripts, summarizes them using gemini-2.5-flash, and allows interactive Q&A about the content. It also integrates OpenAI Whisper for audio transcription if captions are unavailable.

---

Features

• Fetch YouTube video transcripts automatically. reveresed engined - https://youtubetotranscript.com/transcript
• Fallback to Whisper for audio transcription if subtitles are missing.
• Summarize transcripts concisely using gemini-2.5-flash.
• Ask questions interactively about the video content, with AI-powered streaming responses.
• Colorful, user-friendly terminal interface using Rich.

---

Installation

1. Clone the repository:

git clone https://github.com/krishcodes07/TubeInsight.git
cd TubeInsight

2. Install dependencies:

pip install -r requirements.txt

3. Create a .env file in the project root with your Gemini API key:

GEMINI_API_KEY=your_api_key_here

---

Usage

Run the main application:

python app.py

1. Enter the YouTube URL when prompted.
2. The transcript will be fetched. If unavailable, audio will be downloaded and transcribed using Whisper.
3. A summary of the video will be displayed.
4. Ask questions about the video interactively in the terminal.
5. Type exit or quit to stop.

---

Project Structure

TubeInsight/
│
├─ app.py                # Main script
├─ .env                  # Environment variables (API keys)
├─ requirements.txt      # Python dependencies
├─ data/                 # Transcripts saved here
├─ utils/
│  └─ logger.py          # Rich logging setup
├─ services/
│  ├─ youtube.py         # YouTube audio downloader
│  ├─ transcript.py      # Transcript fetching & Whisper transcription
│  └─ gemini.py          # Gemini AI summarization & Q&A
└─ README.md

---

Dependencies

• Rich – For colorful terminal output
• Whisper – Audio transcription
• pytubefix – YouTube video download
• Requests – HTTP requests
• BeautifulSoup4 – HTML parsing
• Google Generative AI – Gemini AI interaction

---

Notes

• Make sure you have a stable internet connection for transcript fetching and Gemini API calls.
• Audio transcription may take longer depending on video length.
• Transcript files are saved in the data/ folder for reuse.

---

License

MIT License © 2025 Developed by Krish