GitHub - pierrepocreau/Marginal-IA: A voice-to-note app removing the friction of taking digital notes while reading physical books

A voice-to-note application for readers. Capture thoughts while reading physical books using voice recordings, and let AI structure them into organized, tagged notes.

Demo

Try it out at marginal-ia.com

Note: The demo runs on free tiers of Railway, Groq, and Supabase, so you may occasionally hit rate limits. For heavy usage, consider self-hosting with your own API keys.

Features

Voice Recording - Record thoughts hands-free while reading
AI-Powered Parsing - Automatically extracts quotes, comments, page numbers, and tags from transcriptions
Book Context Awareness - AI corrects phonetic transcription errors using book/author context
ISBN Lookup - Add books by entering their ISBN
Export Options - Export to Obsidian (markdown with frontmatter) or CSV (Notion, Excel)
Multilingual - Preserves the original language of your notes

Tech Stack

Frontend: Streamlit
Backend: Supabase (PostgreSQL + Auth)
AI: Groq (Whisper for transcription, Llama for parsing)
Book Data: Open Library API

Getting Started

Prerequisites

Python 3.11+
uv package manager
Supabase account
Groq API key

Installation

Clone the repository:

git clone https://github.com/yourusername/marginal-ia.git
cd marginal-ia

Install dependencies:
Set up Supabase:
- Create a new Supabase project
- Run the schema from supabase_schema.sql in the SQL editor
- Enable Row Level Security (RLS) policies are included in the schema

Configure environment variables:

Fill in your credentials:

SUPABASE_URL=your_supabase_url
SUPABASE_KEY=your_supabase_anon_key
GROQ_API_KEY=your_groq_api_key

Run the app:
```
uv run streamlit run main.py
```

Docker

Build the image:
```
docker build -t marginal-ia .
```

Run the container:

docker run -p 8501:8501 \
  -e SUPABASE_URL=your_supabase_url \
  -e SUPABASE_KEY=your_supabase_anon_key \
  -e GROQ_API_KEY=your_groq_api_key \
  marginal-ia

Open http://localhost:8501 in your browser

Deployment on Streamlit Cloud

Push your code to GitHub
Connect your repo to Streamlit Cloud
Add your environment variables in the Streamlit Cloud dashboard
The packages.txt file ensures system dependencies (ffmpeg, libzbar) are installed

Project Structure

marginal-ia/
├── main.py                 # Entry point (auth + routing)
├── pages/
│   ├── login.py            # Authentication
│   ├── recorder.py         # Voice recording & AI parsing
│   ├── books.py            # Book management
│   └── notes.py            # Note viewing & export
├── structures/
│   ├── book.py             # Book dataclass
│   └── note.py             # Note dataclass
├── utils/
│   ├── db.py               # Database client
│   ├── sidebar.py          # Navigation
│   ├── parser.py           # AI note structuring
│   ├── isbn.py             # ISBN lookup
│   └── export.py           # Export functionality
├── supabase_schema.sql     # Database schema
├── pyproject.toml          # Dependencies
└── packages.txt            # System dependencies

Usage

Sign up/Login with email and password
Add a book manually or by entering its ISBN
Select the book from the sidebar
Record your thoughts while reading
Review the AI-parsed note (quote, comment, tags)
Export your notes to Obsidian or CSV when ready

Note Tags

The AI automatically assigns one or more tags to each note:

character - Character analysis or observations
question - Questions raised by the text
remark - General remarks
quote - Notable quotes worth remembering
summary - Chapter or section summaries
idea - Ideas sparked by the reading
connection - Connections to other works or concepts
critique - Critical analysis

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

This project is licensed under the MIT License - see the LICENSE file for details.