Chitti Notes AI

Inspiration

Students attend hours of lectures every day, but taking notes while listening to a teacher can be difficult. Important points can be missed, and manually organizing notes after class takes a lot of time.

This inspired us to build Chitti Notes AI — an AI-powered tool that turns recorded lectures into clear, structured, and useful study notes.

What it does

Chitti Notes AI allows students to record or upload a lecture and automatically transform the audio into organized notes.

The system:

• Converts lecture audio into text using speech-to-text technology.
• Processes the transcript using AI.
• Generates concise and structured study notes.
• Helps students quickly revise important concepts without going through the entire recording.

How we built it

The application was built using a combination of modern AI and software technologies.

Frontend: Flutter Backend: Python + FastAPI AI: Speech-to-text + Large Language Model LLM: Ollama APIs: Gemini API Communication: REST API

The Flutter application records the lecture audio and sends it to the FastAPI backend. The backend processes the audio, converts it into a transcript, and passes the transcript to an AI model to generate structured notes. The generated notes are then returned to the application for the student to read and revise.

Challenges we faced

One of the major challenges was processing lecture audio efficiently and converting long conversations into meaningful notes. We also had to handle communication between the Flutter frontend and Python backend, API integration, CORS issues, and AI model resource limitations.

Another challenge was making the generated notes useful for students rather than simply producing a large block of text. We improved the prompting and output structure so that the AI generates more readable and revision-friendly notes.

What we learned

Through this project, we learned how to integrate AI models with a real-world application, build APIs using FastAPI, connect a Flutter frontend with a Python backend, process audio, and design prompts for generating useful AI outputs.

Most importantly, we learned that building an AI application is not only about using an AI model — the real challenge is creating a reliable workflow around it that solves an actual user problem.

Future improvements

We plan to improve Chitti Notes AI with features such as:

• Automatic key-point extraction
• Important-question generation
• Flashcards and quizzes
• Multi-language support
• Speaker identification
• Searchable lecture history
• Personalized study plans

Chitti Notes AI aims to make lecture revision faster, smarter, and more accessible for students.

Built With

• artificial
• fastapi
• intelligence
• python