YouTube Assistant

A Streamlit web app to extract, summarize, translate, and interact with YouTube video transcripts using Gemini AI.
(Text-to-Voice feature coming soon!)

Features

• Extract YouTube Transcripts

  • Enter any YouTube URL and fetch the transcript.
  • Automatically grabs in the first available language.
  • Handles long transcripts by chunking & summarizing in parts.

• Summarize Transcript

  • Get concise, bullet-point summaries powered by Gemini AI.

• View Transcript

  • Access the complete transcript in its original language directly from the sidebar.

• Ask Questions

  • Interactively ask about the video content.
  • Receive AI-powered answers grounded in the transcript.

• Multilingual Support

  • Translate summaries and answers into 20+ languages via Gemini.
  • UI elements (buttons, headings, prompts) are instantly translated with deep-translator.

• Interactive UI

  • Clean Streamlit interface with spinners, sidebars, and buttons for smooth interaction.

• Text-to-Voice (Coming Soon)

  • Convert transcript summaries and answers into speech.

Tech Stack

• Streamlit: For building the interactive web frontend.
• youtube-transcript-api: To fetch YouTube video transcripts programmatically.
• google-generativeai (Gemini 2.0 Flash-Lite model): For summarization, translation, and question answering on video content. Model, rate, and token limits are configurable in config.py.
• deep-translator: For translating UI elements content into multiple languages.
• python-dotenv: Managing API keys securely.

Setup

1. Clone the repository:

   git clone https://github.com/SaiSruthisri/YoutubeAssistant.git
   cd YoutubeAssistant

2. Set up API keys:

   • Create a .env file in the project root.
   • Add your Gemini API key:

     GEMINI_API_KEY=your_gemini_api_key_here
     

3. Install dependencies:

   pip install -r requirements.txt

4. Run the app:

   streamlit run app.py

   • After running, open your browser and go to http://localhost:8501 to use the app live.

Usage

1. Choose Language:
   Select your preferred language for the UI and output.

2. Enter YouTube URL:
   Paste a valid YouTube video URL in the input box.

3. Load Transcript:
   Click "Load Transcript" to fetch and summarize the transcript.

4. Summarize the Video:
   Click "📝Summarize the video" for a bullet-point summary.

5. View Entire Transcript:
   After loading, you can view the complete transcript in its original language from the sidebar.

6. Ask Questions:
   Type your question about the video and click "Ask" to get an AI-powered answer.

7. Translate Transcript, Summary, and Answers:

   • Summaries and answers are generated in English by Gemini, then translated to your chosen language using Gemini for high quality.
   • UI elements (headings, buttons, prompts) are translated instantly using deep-translator for a native experience.

8. Text to Voice:
   Feature coming soon !

Note: Since this is a Streamlit app, always use the buttons in the intended order (e.g., Load Transcript → Summarize → Ask) to avoid losing data on the screen. Switching between buttons or reloading may reset the app state and clear your previous results or questions.

Supported Languages (for translation)

• English, Hindi, French, Spanish, German, Telugu, Kannada, Tamil, Russian, Portuguese, Italian, Dutch, Arabic, Korean, Turkish, Vietnamese, Filipino, Japanese, Chinese

Special Thanks & Documentation

This project was made possible thanks to the following open-source libraries, documentation, and best practices:

• Streamlit
• youtube-transcript-api
• deep-translator
• google-generativeai
• python-dotenv
• Gemini API Rate Limiting: Read about Gemini model rate limits and quotas here
• Retry Strategies: Best practices for handling API retries and backoff are explained here

Working flow

youtubeAssistant.mp4