Loading repository data…
Loading repository data…
Ebrizzzz / repository
A desktop application that extracts YouTube playlist transcripts and enhances them using Google's Gemini AI models. The output is a book in any language you want.
A transparent discovery signal based on current public GitHub metadata.
This score does not audit code, security, maintainers, documentation quality, or suitability. Verify the repository and its current documentation before adoption.
✅ Added several Refinement styles to choose from based on your specific needs.
The "Refinement Style" dropdown allows you to choose how AI will process the YouTube transcript. Here's a description of each style:
⚖️ Balanced and Detailed: This is the default style, providing a comprehensive refinement of the transcript. It focuses on organizing the text into a well-structured, readable format with headings, bullet points, and bold text, while preserving every detail, context, and nuance of the original content. Ideal if you want a thoroughly enhanced transcript without any information loss.
📝 Summary: This style generates a concise and informative summary of the video transcript. It extracts the core message, main arguments, and key information, providing a quick and easily digestible overview of the video's content. Best for when you need to quickly grasp the main points without reading the entire transcript.
📚 Educational: This style transforms the transcript into a structured educational text, similar to a textbook chapter. It uses headings, subheadings, and bullet points for clarity and organization, making it ideal for learning. Crucially, it also identifies and defines technical terms and jargon within blockquotes, enhancing understanding and acting as a built-in glossary. (Example Image Below)
✍️ Narrative Rewriting: This style creatively rewrites the transcript into an engaging narrative or story format. It transforms the factual or conversational content into a more captivating and readable piece, like a short story or narrative article. While storytelling is applied, it stays closely aligned with the original video's subjects and information, making the content more accessible and enjoyable.
❓ Q&A Generation: This style generates a set of questions and answers based on the transcript, formatted for self-assessment or review. Each question is presented as a foldable header (using Markdown), with the answer hidden beneath. This format is perfect for creating study guides or quizzes to test your understanding of the video content.(Example Image Below)
✅ Added Language Support, now the output file is in the language of user's input. Also, the video can be in any language, as long as a transcript exists, it will work. ✅ Added single video url support, no need to put it in a playlist. ✅ Added configurable Chunk Size for API calls.
Users can now control the chunk size used when processing transcripts with the Gemini API via a slider in the UI. This allows for customization of processing behavior:
- Larger chunk sizes: Reduce the number of API calls, potentially speeding up execution and suitable for summarizing longer videos with less emphasis on fine details.
- Smaller chunk sizes: Increase API calls but may preserve more detail and nuance, potentially beneficial for tasks requiring high fidelity output. ❓ What is Chunk Size? A video, is divided into chunks to be given to AI, so if you set chunk size to 3000 words, and the video has 8000 words, the API workflow would be like this :
- First 3000 words ➡➡processed by AI➡➡ Refined part 1
- Second 3000 words + Refined part 1 as context ➡➡processed by AI➡➡ Refinde part 2
- final 2000 words + Refined part 1 + 2 as context ➡➡processed by AI➡➡ Refinde part 3
- Refined part 1 + Refined part 2 + Refined part 3 = Final Formatted Text of the video!
Read more about it in this Medium Article
pip install -r requirements.txt
python main.py
Example of Educational Style with added definition of technical terms Example of Q&A Style, Questions are headers so they can be folded/unfolded
YouTube playlist used for example files : https://www.youtube.com/playlist?list=PLmHVyfmcRKyx1KSoobwukzf1Nf-Y97Rw0