Study-Budy

Inspiration

As students, we struggled with dense textbooks and last-minute exam prep. Traditional study tools felt passive, while AI advancements offered untapped potential. We envisioned StudyBuddy—a tool that bridges the gap by making documents listenable and summarizeable with AI, catering to auditory learners and busy students.

Document Processing

Converts PDFs, Word files, and text into audio lectures with clear pronunciation.

Extracts key concepts from study materials in seconds.

Supports dyslexic and auditory learners with customizable playback speeds.

Smart Features

Highlight-based summary generation.

Keyboard shortcuts for accessibility.

Cross-platform compatibility (web + mobile-responsive)

How we built it

Frontend: React.js + Vite (for blazing-fast builds)
Document Processing: pdf-lib (PDFs), mammoth (DOCX)
Audio: Web Speech API (TTS) + custom pronunciation dictionaries
Summarization: Rule-based extraction + Hugging Face transformers (for key points)

Challenges I'm ran into

PDF Text Extraction: Formatting nightmares! We solved it with positional text mapping.

Real-Time TTS: Avoiding robotic delivery required chunking text and adding pauses.

Cross-Browser Support: Firefox hated our audio code. Polyfills to the rescue!

What we learned

Document Processing: Extracting clean text from PDFs/DOCs (hello, pdf-lib and mammoth!)

AI Limitations: Balancing accuracy with speed in summarization (we opted for hybrid rule-based + NLP approaches)

Accessibility: Designing for screen readers and keyboard navigation wasn’t optional—it was essential.

What's next for Study-Budy

Enhanced NLP Engine:

(Current limitation: Works best with documents under 20 pages)

Scaling to handle textbooks/research papers with chapter-wise analysis Implementing BERT-based summarization for more accurate key points Adding citation extraction for academic papers

Gamified Learning:

"Quiz Attack" mode: AI generates flashcards from your docs Study Streaks: Earn points for daily learning Knowledge Battles: Challenge friends on extracted concepts