Turn lectures into slides, quizzes, and study material automatically.
NoteAI is an iOS app that converts recorded or imported audio into structured study material using on-device speech recognition and AI-powered analysis.
Record a lecture → generate slides → take a quiz → study faster.
NoteAI is designed to help students transform long lectures into structured study resources.
Instead of manually summarizing notes, NoteAI automatically generates:
- 📑 Slide outlines
- 🧠 Quiz questions
- 📝 Structured transcripts
The app combines local speech-to-text models with AI analysis to turn raw lecture audio into organized learning material.
Everything runs inside a native SwiftUI iOS app designed for a clean, focused study experience.
- Record lectures directly in the app
- Import existing audio recordings
- On-device speech recognition using Whisper-based models
- No external transcription service required
- Save multiple transcripts
- Rename transcripts
- Delete transcripts
- Clean library UI for browsing notes
Generate structured study material from transcripts:
- 📑 Slide outlines
- ❓ Quiz questions
- 🧠 Context-aware lecture interpretation
- Swipeable slide cards
- Interactive quiz interface
- Quiz scoring and retry
- Clean, distraction-free UI
- Fully native SwiftUI design
- Custom design system
- Light & dark mode support
- Consistent card-based layout
| Home | Transcript | Slides | Quiz |
|---|---|---|---|
 |
 |
 |
 |
Language
Swift
Frameworks
SwiftUI
AudioKit
SwiftWhisper
AI
Google Gemini API
Architecture
- Local speech recognition
- AI content generation pipeline
- Local transcript persistence
Audio Input
↓
Local Speech Recognition (Whisper)
↓
Transcript Storage
↓
Gemini AI Processing
↓
Slides + Quiz Generation
↓
SwiftUI Study Interface
1️⃣ Clone the repository git clone https://github.com/11samm/NoteAI.git
2️⃣ Open the project in Xcode
3️⃣ Add your Gemini API key: NoteAI/Config/APIKey.swift
4️⃣ Build and run on simulator or device
Speech recognition runs locally using Whisper-based models so transcription works without external APIs to ensure privacy and be efficient with costs.
On-device transcription can contain errors.
Prompts are designed so Gemini:
- uses contextual clues
- corrects likely transcription mistakes
- avoids hallucinating missing information.
Recorded audio is in the .m4a format, but many recordings are .mp3 so a formatter had to be added before it is processed for transcription
Planned improvements:
- 📚 Long lecture support (3+ hour transcripts)
- 🧠 Flashcard generation
- 🔍 Transcript search
- 📤 Export notes
- 🎧 YouTube/Zoom import
Samuel Garcia
Computer Science student focused on building AI-powered tools and mobile apps.
GitHub
https://github.com/11samm
Give it a star on GitHub ⭐
