Notebolt: Your AI Co-Pilot for Effortless Lecture Notes

• 

  Load the Replicate Key and Upload Lecture Audio

• 

  Continuation of the Note

• 

  Lecture Audio is Transcribe and Snowflake's Artic Instruct Model Generates a Structured Notes in context with the lecture

• 

  Ask further questions about the Notes Generated and Snowflake Artic Instruct Model gives you an answer in context with the lecture notes

Inspiration

Remember those times in class when you were scribbling notes so fast your hand cramped, but you still missed half of what the teacher said? Notebolt was born out of that struggle! It's an AI-powered note-taking assistant designed to help students capture key ideas from lectures without the stress.

What It Does

Notebolt acts as your personal co-pilot in the classroom. It transcribes and summarizes lecture audio into clear, structured notes: highlighting important points and summarizing key concepts. This frees you to focus on understanding the material, participating in discussions, and asking questions. With our interactive chat interface, students can ask questions, get clarifications, and delve deeper into specific topics, making studying easier and more effective.

How We Built It

Notebolt was built using a combination of cutting-edge technologies:

• AI Transcription: We used Whisper, an advanced speech recognition model, to accurately transcribe lecture audio.
• AI Summarization: Snowflake Artic LLM was utilized to generate clear and concise summaries of the transcribed text.
• Interactive Q&A: Just like talking to a friend, we integrated a chat interface to allow students to ask questions and get clarifications on the generated notes.
• Streamlit: The user interface was built with Streamlit, providing an easy-to-use platform for students to interact with Notebolt.

Challenges We Ran Into

• Accuracy of Transcriptions: Ensuring Whisper base model could accurately transcribes various accents and lecture environments was a significant challenge.
• Summarization Quality: Getting Snowflake's Artic instruct model to be able to effectively summarize long lectures into concise notes required extensive prompting and validation.
• User Interface Design: Designing a simple but intuitive and accessible interface using Streamlit that caters to diverse user needs involved iterative design and user testing.

Accomplishments That We're Proud Of

• Successfully integrating multiple AI models (Whisper and Snowflake's Artic instruct model) to create a seamless and functional application.
• Developing an intuitive user interface that caters to diverse learning styles.
• Creating a tool that has the potential to significantly improve students' learning experiences by reducing the stress of note-taking.

What We Learned

Throughout the development of Notebolt, we learned about the complexities of natural language processing, the importance of user-friendly design, and the diverse needs of students. We also gained valuable experience in integrating various technologies to create a seamless user experience.

What's Next for Notebolt: Your AI Co-Pilot for Effortless Lecture Notes

• Search Function: Allow students to search for specific keywords within the transcribed text or summaries for easy review.
• Customizable Notes: Let users personalize the level of detail in the summary (bullet points, short paragraphs) based on their learning style.
• Flashcards & Quizzes: Generate flashcards and quizzes based on key points to help students test their understanding and prepare for exams.

Built With

• fpdf
• github
• python
• replicate
• snowflake-llm
• streamlit
• torchaudio
• transformers
• whisper
• youtube