RecipeXtractor

Inspiration

Nowadays, there are so many nice cafes and comfort food is everywhere—but it’s often expensive. We wanted to make it easy for anyone to enjoy great meals at home without spending hours searching. Instead of watching long videos to figure out ingredients and steps, RecipeXtractor gives you a clear grocery list, exact quantities, and instructions just minutes after you paste a YouTube link. Cooking becomes easy, fast, and fun—so you can spend more time enjoying your food.

What it does

RecipeXtractor takes a YouTube recipe video link, extracts the transcript and instructions, and generates a complete grocery list with quantities. It summarizes all the information clearly, so you don’t have to watch hours of video or guess what to buy. You can start cooking within minutes.

How we built it

We built the app using Kotlin for the backend and integrated Whisper to generate transcripts locally. We use Gemini for smart summarization and context awareness. Right now, the backend runs locally—we haven’t deployed it yet. The workflow is:

Take YouTube link Extract video audio Use Whisper to generate transcript Summarize using Gemini Output grocery list + step-by-step instructions.

Challenges we ran into

The biggest challenge was working with Whisper and integrating Gemini seamlessly. Initially, we tried to use YouTube transcripts directly, but most videos block access. Switching to Whisper for audio transcription was a key decision, but integrating it smoothly with the summarization pipeline took time.

Accomplishments that we're proud of

We built something meaningful, functional, and really easy to use. Unlike Instagram suggestions that just show you food, RecipeXtractor actually lets you cook it at home. It’s fast, intuitive, and solves a real problem for anyone who loves cooking but hates watching endless videos.

What we learned

We learned a lot about audio transcription, summarization, and context-aware AI integration. Beyond the technical side, we realized the importance of user experience—giving users exactly what they need in a few minutes, not hours of content.

What's next for RecipeXtractor

Next, we want to add translation support, since recipes exist in many languages worldwide. We also plan to use AI prompts to suggest alternatives for allergies, dietary restrictions, or ingredient substitutions. Eventually, the app could become a universal cooking companion for any cuisine around the world.

Built With

• android-studio
• flask
• kotlin
• localhost
• python
• whisper
• ytdls