📚 Accessible Voice-Based Course Advisor

A voice-navigated chatbot that helps students, especially from the disability community, find and register for required university courses quickly, easily, and entirely through speech.

💡 Inspiration

Course registration is often overwhelming, visually demanding, and difficult to navigate, especially for students with visual, motor, or learning disabilities. We wanted to create a tool that empowers these student by making the process fully accessible through conversational voice interactions.

🧠 What It Does

Enroll Insight is an accessibility focused application that utilizes speech to text and text to speech technologies alongside an AI assistant to make course registration/schedule creation at NJIT more inclusive.

Accepts voice input from the user
Asks follow-up questions about academic year, time preferences, and career goals
Recommends personalized courses
Responds via voice using text-to-speech
Avoids recommending the same course more than once
Keeps context using session tracking

🛠️ How We Built It

Frontend: React + Tailwind CSS
Speech Processing: Azure Cognitive Speech Services (STT + TTS)
Backend: FastAPI + Python
AI Chat Layer: Gemini AI (Google GenAI)
Database: MongoDB for storing synced NJIT course data
Course Sync: Custom API client that fetches latest courses from NJIT and syncs them to MongoDB

🚧 Challenges We Faced

Gemini was inconsistent: ignored instructions and often repeated recommendations
Prompt tuning required many iterations
We had to build a recommendation tracker to prevent duplicates

🏆 Accomplishments

Built a full accessibility-focused tool in under 24 hours
Created real-time voice-first UX
Improved accessibility and inclusion for a real use case

🎓 What we learned

We learned how to use multiple new technologies during the course of this project.

How to utilize Microsoft Azure's speech resources to integrate text to speech and speech to text into the project
How to tailor prompts for Google Gemini in order to get the most accurate results
How to create a frontend using React and a backend using FastAPI + Python