H5P Interactive Video Generator (WIP) 🎥✨

Welcome to the H5P Interactive Video Generator! This tool transforms YouTube videos into interactive H5P lessons for Moodle using Gemini AI. Built with Streamlit, it’s a work-in-progress that’s already functional but still being polished. Here’s what it does, what works, and how you can get started!

---

🌐 External Resources

Copy the link, as GitHub is not letting it hot link out to my acs.h5p hub

• Sample output and import into H5P.com hub

---

🚀 Features

• Video Summary: Turns a YouTube video transcript into a concise summary (100-150 words) with timestamps (e.g., "Weekly Team Meetings" at 00:50).
• Interactive Activities: Generates activities like multiple-choice questions (MCQs) from the summary (e.g., "What’s the purpose of the 'parking lot'?" at 03:30).
• H5P Output: Packages the video and interactions into an H5P file for Moodle, plus a Markdown file with questions.
• User Interface: A Streamlit app lets you input a URL, pick a learning outcome (e.g., "summary", "analysis"), and download results.

✅ What Works

• Summary Generation: Reliably creates accurate summaries from YouTube transcripts using Gemini AI. For example, https://youtu.be/xv49RRuAI7w gets a spot-on meeting structure summary.
• H5P Packaging: Successfully outputs H5P files and Markdown question lists, ready for Moodle.
• Streamlit Interface: Easy-to-use UI for inputting URLs and downloading files.

🛠️ What Needs Work

• Interaction Generation: Sometimes generates off-topic questions (e.g., narrative themes instead of meetings). We’re debugging the prompt replacement to lock it to the summary.
• Timestamp Alignment: Fallback questions work but don’t always match summary timestamps—aiming for dynamic alignment.
• Prompt Selector: Currently basic; plans to enhance it with summary keyword matching.

🏁 Quick Start (For Python Newbies)

Don’t worry if you’re new to Python—we’ve got you covered with a step-by-step setup!

Prerequisites

• Python 3.9+: Download from python.org. Check with python --version in your terminal.
• Git: Install from git-scm.com to clone this repo.
• A Terminal: Use Command Prompt (Windows), Terminal (macOS/Linux), or an IDE like VS Code.

Setup Process

1. Clone the Repo:

   git clone https://github.com/yourusername/h5p-interactive-video.git
   cd h5p-interactive-video

   Replace yourusername with your GitHub username after uploading.

2. Create a Virtual Environment (keeps dependencies tidy):

   python -m venv venv

   • Windows: venv\Scripts\activate
   • macOS/Linux: source venv/bin/activate You’ll see (venv) in your terminal—nice!

3. Install Dependencies:

   pip install streamlit google-generativeai youtube_transcript_api

   If pip fails, try python -m pip install ....

4. Get a Gemini API Key:

   • Visit Google AI Studio.
   • Sign in, create a key, and copy it.

5. Run the App:

   streamlit run app.py

   • A browser window should pop up. Enter your YouTube URL (e.g., https://youtu.be/xv49RRuAI7w) and API key, then hit "Generate".

Example Usage

• Input: https://youtu.be/xv49RRuAI7w, select "analysis".
• Output: Downloads an H5P file and Markdown with questions (e.g., "What does the red/yellow/green approach track?").

⚙️ How It Works

1. 📥 Input: Enter a YouTube URL + Gemini API key.
2. 📜 Transcript Fetch: Pulls the transcript with youtube_transcript_api.
3. 🤖 Summary Generation: Gemini AI crafts a summary with timestamps.
4. ❓ Interaction Generation: Creates activities from the summary (WIP for consistency).
5. 📦 Packaging: Outputs H5P and Markdown files.

🐛 Known Issues

• Off-Topic Questions: Interaction generation may drift (e.g., narrative instead of meetings). Fix in progress.
• Fallback Reliance: Uses hardcoded questions if Gemini fails, but timestamps may not align perfectly.

🌟 Contributing

This is a WIP, and we’d love your help! Fork the repo, tweak the code, and submit a PR. Focus areas:

• Fixing generate_interactions prompt replacement.
• Enhancing prompt_selector for keyword-based logic.

📜 License

MIT License—feel free to use and modify!

🙏 Thanks

Built with ❤️ by ACS and the xAI Grok assistant.