NoTeeline: Supporting Real-Time, Personalized Notetaking with LLM-Enhanced Micronotes

Write, Organize, Review, and Summarize Personalized Notes

Website • Doi • Arxiv • Huggingface

✨ Teaser

noteeline.mp4

You can also find the long preview in YouTube for a better quality.

📰 News

• [13-09-2026] This fork now runs fully self-hosted: it transcribes a live in-person lecture from your microphone with a local Whisper model instead of pulling a YouTube transcript, and expands micronotes with a local Ollama model instead of the OpenAI API. Nothing leaves your machine, and no API key is needed.
• [13-09-2026] The whole thing runs from one command with Docker — see Running it (Docker) below.
• [28-01-2025] The micronotes can be expanded using the OpenAI API which produces streamlined real-time output
• [28-01-2025] A personal OpenAI API key is to be provided on the landing page of NoTeeline for the app to work
• [28-01-2025] NoTeeline got featured in Huggingface, link

🐳 Running it (Docker)

This is the easiest way to run the app — one command starts the frontend, backend, a local Whisper transcription server, and a local Ollama LLM together:

$ docker compose up -d

Then open http://localhost:3000. The first run pulls the LLM and Whisper models in the background (a few GB total), so give it a few minutes the first time — the app itself is usable immediately while that finishes. Everything after that is instant, since the models are cached in Docker volumes.

Click Start Recording Lecture, let it listen, and type short keypoints as you go — they'll be expanded into full notes using your local model and the live transcript around the moment you typed them.

🛠️ Local Development

If you'd rather run the pieces yourself instead of through Docker:

• Clone the repo using the following command: $ git clone https://github.com/Xopher00/NoTeeline.git
• If npm is the desired package manager, then just replace the command:
  • yarn with the command npm install
  • yarn dev with the command npm run dev

You'll also need Ollama running locally with a model pulled (e.g. ollama pull llama3.1:8b), and a local Whisper server (Backend/whisper_server.py, a small FastAPI app built on faster-whisper) running on port 8000.

💻 Frontend

• cd into the folder Frontend Run yarn to install all the necessary packages
• While the current working directory is Frontend, run the command yarn dev to start the frontend

⚙️ Backend

• To start the server, cd into the Backend folder and run yarn to install all the necessary packages for the backend server
• While in the Backend directory, run the command node index.js to start the server at port 4000
• In a separate terminal, set up a Python virtual environment in Backend and install faster-whisper fastapi uvicorn python-multipart, then run uvicorn whisper_server:app --host 0.0.0.0 --port 8000 to start the transcription server

📜 Citation

If you use the NoTeeline code or data, please cite our paper:

@inproceedings{faria2025noteeline,
  author = {Faria Huq, Abdus Samee, David Chuan-En Lin, Alice Xiaodi Tang, and Jeffrey P Bigham},
  title = {NoTeeline: Supporting Real-Time, Personalized Notetaking with LLM-Enhanced Micronotes},
  year = {2025},
  publisher = {Association for Computing Machinery},
  address = {New York, NY, USA},
  booktitle = {Proceedings of the 30th International Conference on Intelligent User Interfaces},
  numpages = {18},
  location = {Cagliari, Italy}
}

🌟 Questions?

Please feel free to email Faria (fhuq@cs.cmu.edu)/ Samee (abdussamee16@gmail.com). Alternatively, you can open a new issue over github!