Rec2Note

WIP - Active development

Rec2Note is a CLI tool for transforming lecture recordings into student notes that capture key concepts delivered with visual aids (slides, whiteboard, …), coursework-related notices, and questions from students.

Current State

The app is in active development. Currently available:

• CLI interface for processing transcripts
• Minimal pipeline (summary generation)
• Full pipeline (summary + deadlines + study questions + student Q&A)
• Rich terminal output with preview

Planned features (not yet implemented):

• Streamlit web UI
• Textual TUI
• Auto-transcription from media files
• SQLite database for storing notes

Features:

• Generate detailed summaries (title, key points, topics, key terms) from transcripts
• Extract important deadlines (exam dates, assignment due dates, …)
• Extract student questions and answers
• Generate study questions
• Optional: include media file path for future visual-aid timestamp extraction
• Preview generated notes directly in terminal

Prerequisites

• Python 3.13+
• uv (recommended) or pip to install dependencies
• Google Gemini API key — create a .env file in the project root with:

  GOOGLE_GEMINI_API_KEY=your_api_key_here

  (The app loads this via pydantic-settings; see config.)

No build step is required: install deps and run. The SQLite DB and data/ directory are created automatically on first use.

Running the app

Install dependencies, then run the CLI:

# Install the package in development mode
uv pip install -e .

# Process a transcript with minimal pipeline (summary only)
rec2note process run -tr path/to/transcript.txt

# Process with full pipeline and preview output
rec2note process run -tr path/to/transcript.txt --full --preview

# Include media file for reference
rec2note process run -media path/to/lecture.mp4 -tr path/to/transcript.txt --full

# Custom output directory
rec2note process run -tr path/to/transcript.txt -o ./my-notes

CLI Options

Option	Short	Description
--transcript	-tr	Path to transcript file (.txt or .md) — required
--media	-media	Path to audio/video file (optional, for metadata)
--minimal	-m	Run minimal pipeline (summary only) — default
--full	-f	Run full pipeline (summary + deadlines + study questions + student Q&A)
--preview	-pr	Render generated markdown in terminal
--output-dir	-o	Output directory for generated notes (default: ./notes)

Current limitations

• Auto-transcription from media is not yet supported — a transcript file is required.
• Streamlit UI and TUI are planned but not yet implemented.

Directory

Rec2Note/
├── .env                    # Secrets (API keys)
├── .gitignore
├── pyproject.toml          # Dependencies (uv)
├── README.md
├── data/                   # SQLite DB (rec2note.db), gitignored
├── recordings/             # Input recordings
│   └── .gitkeep
├── notes/                  # Output notes (gitignored)
│   └── .gitkeep
├── tests/
│   └── ...
└── src/
    └── rec2note_cli/       # Main package
        ├── __init__.py
        ├── main.py         # Entry point (Typer CLI)
        ├── config.py       # Configuration (Pydantic Settings)
        ├── cli/            # CLI commands (process run)
        ├── core/           # Business logic
        │   ├── agents.py   # LLM agents (summary, deadlines, etc.)
        │   ├── db.py       # SQLite storage for lecture notes
        │   ├── llm.py      # Gemini API client
        │   ├── models.py   # Pydantic models and LectureNoteResult
        │   └── pipeline.py # Orchestrates transcript → agents → DB
        ├── enums/
        ├── prompts/        # LLM instruction files
        ├── ui/             # Rich console UI
        └── utils/          # read_file, timestamp (HH:MM:SS → seconds)

Stack

• Package manager: uv
• CLI: Typer + Rich
• LLM: Gemini API
• DB: SQLite (built-in, data/rec2note.db)